Print

Quality of Service

Quality of Service Conceptual Overview

QoS processes apply to traffic that flows through Ethernet ports and control planes. These processes can modify data fields (CoS or DSCP) or assign data streams to traffic classes for prioritized handling. Transmission queues are configurable for individual Ethernet ports to shape traffic based on its traffic class. Many switches also support traffic policies that apply to data that is filtered by access control lists.

The following sections describe QoS features:

QoS Data Fields and Traffic Classes

Quality of Service defines a method of differentiating data streams to provide varying levels of service to the different streams.

Criteria determining a packet’s priority level include packet field contents and the port where data packets are received. QoS settings are translated into traffic classes, which are then used by switches to manage all traffic flows. Traffic flow management varies with each switch platform.

QoS Data Fields

Quality of service decisions are based on the contents of the following packet fields:

  • CoS (three bits): Class of service (CoS) is a 3-bit field in Ethernet frame headers using VLAN tagging. The field specifies a priority value between zero and seven. Class of service operates at layer 2.
  • DSCP (six bits): Differentiated Service Code Point (DSCP) is a 6-bit field in the Type Of Service (TOS) field of IP packet headers.

Port Settings – Trust Mode and Traffic Class

Ethernet and port channel interfaces support three QoS trust modes:

  • CoS Trust: Ports use inbound packet CoS field contents to derive the traffic class.
  • DSCP Trust: Ports use inbound packets DSCP field contents to derive the traffic class.
  • Untrusted: Ports use their default values to derive the traffic class, ignoring packet contents.

The default mode setting is CoS trust for switched ports and DSCP trust for routed ports.

Ports are associated with default CoS, DSCP, and traffic class settings; defaults vary by platform.

These sections describe procedures for configuring port settings:

Rewriting CoS and DSCP

CoS Rewrite

Switches can rewrite the CoS field for outbound tagged packets. The new CoS value is configurable, and is derived from a data stream’s traffic class as specified by the traffic class-to-CoS rewrite map. CoS rewrite is disabled on all the traffic received on CoS trusted ports.

On Arad, Jericho, FM6000, Trident and Tomahawk, Trident II, and Helix platform switches, CoS rewrite can be enabled or disabled on DSCP trusted ports and untrusted ports.

  • CoS rewrite is globally enabled by default for packets received on untrusted ports and DSCP trusted ports if at least one port is explicitly configured in DSCP trust or untrusted mode.
  • CoS rewrite is globally disabled by default for packets received on untrusted ports and DSCP trusted ports if there are no ports explicitly configured in DSCP trust or untrusted mode.

On Petra platform switches, CoS rewrite is always enabled on DSCP trusted ports and untrusted ports.

DSCP Rewrite

Switches can rewrite the DSCP field for outbound IP packets. On FM6000, Trident and Tomahawk, Trident II, and Helix platform switches, DSCP rewrite is disabled by default on all ports and always disabled for traffic received on DSCP trusted ports. On Petra, Arad, and Jericho platform switches, DSCP rewrite is always disabled.

FM6000, Trident and Tomahawk, Trident II, and Helix platform switches provide a command that enables or disables DSCP rewrite for packets received on CoS trusted ports and untrusted ports. The new DSCP value is configurable, based on the data stream’s traffic class, as specified by the traffic class-to-DSCP rewrite map.

These sections describe procedures for rewriting CoS and DSCP fields:

Traffic Classes

Data stream distribution is based on their traffic classes. Data stream management varies by switch platform. Traffic classes are derived from these data stream, inbound port, and switch attributes:

  • CoS field contents
  • DSCP field contents
  • Inbound port trust setting
  • CoS default setting (Arad, Jericho, FM6000, Trident and Tomahawk, Trident II, and Helix platform switches)
  • DSCP default setting (Arad, Jericho, FM6000, Trident and Tomahawk, and Trident II platform switches)
  • Traffic class default setting (Petra platform switches)

When a port is configured to derive a data stream’s traffic class from the CoS or DSCP value associated with the stream, the traffic class is determined from a conversion map.

  • A CoS-to-traffic class map derives a traffic class from a CoS value.
  • A DSCP-to-traffic class map derives a traffic class from a DSCP value.

Map entries are configurable through CLI commands. Default maps determine the traffic class value when CLI map entry commands are not configured. Default maps vary by switch platform.

These sections describe traffic class configuration procedures:

Transmit Queues and Port Shaping

Transmit queues are logical partitions of an Ethernet port’s egress bandwidth. Data streams are assigned to queues based on their traffic class, then sent as scheduled by port and transmit settings. Support varies by switch platform. A queue’s label determines its priority: queues with the suffix “0” have the lowest priority.

Parameters that determine transmission schedules include:

  • Traffic class-to-transmit queue mapping determines the transmit queue for transmitting data streams based on traffic class. The set of available transmit maps vary by switch platforms:
    • Arad, Jericho, FM6000, Trident II, and Helix platforms: one map for all unicast and multicast traffic.
    • Trident and Tomahawk platform: one map for unicast traffic and one map for multicast traffic.
    • Petra platform: one map for unicast traffic. Queue shaping is not available for multicast traffic.
  • Port shaping specifies a port’s maximum egress bandwidth.
  • Queue shaping specifies a transmit queue’s maximum egress bandwidth, and implementation varies by platform.
    • Trident and Tomahawk platform: queue shaping is configurable separately for unicast and multicast queues.
    • Trident II platform: queue shaping is configurable for transmit queues. Port shaping and queue shaping are supported only in store-and-forward switching mode.
    • Petra platform: queue shaping is not available for multicast traffic.
    • Helix platform: queue shaping is configurable for transmit queues.
    • FM6000 platform: switches do not support simultaneous port shaping and queue shaping. Enabling port shaping on an FM6000 switch disables queue shaping, regardless of the previous configuration.
  • Guaranteed bandwidth guarantees the allocation of a specified bandwidth for a transmit queue. Guaranteed bandwidth is supported only on Trident II platforms.
  • Queue priority specifies the priority at which a transmit queue is serviced. The switch defines two queue priority types:
    • Strict priority queues are serviced in the order of their priority rank - subject to each queue’s configured maximum bandwidth. Data is not handled for a queue until all queues with higher priority are emptied or their transmission limit is reached. These queues typically carry low latency real time traffic and require highest available priority.
      • Round robin queues are serviced simultaneously subject to assigned bandwidth percentage and configured maximum bandwidth. All round robin queues have lower priority than strict priority queues. Round robin queues can be starved by strict priority queues.
  • Queue scheduling determines how packets from different transmit queues are serviced to be sent out on the port.
  • Queue bandwidth allocation specifies the time slice (percentage) assigned to a round robin queue, relative to all other round robin queues.

These sections describe transmit queue and port shaping configuration procedures:

Explicit Congestion Notification (ECN)

Explicit Congestion Notification (ECN) is an IP and TCP extension that facilitates end-to-end network congestion notification without dropping packets. ECN recognizes early congestion and sets flags that signal affected hosts. Trident and Tomahawk, Trident II, and Helix platform switches extend ECN support to non-TCP packets.

ECN usage requires that it is supported and enabled by both endpoints. Although only unicast flows are modified by ECN markers, the multicast, broadcast, and unmarked unicast flows can affect network congestion and influence the indication of unicast packet congestion.

ECN Conceptual Overview

The ECN field in the IP header (bits 6 and 7 in the IPv4 TOS or IPv6 traffic class octet) advertises ECN capabilities:

  • 00: Router does not support ECN.
  • 10: Router supports ECN.
  • 01: Router supports ECN.
  • 11: Congestion encountered.

Networks typically signal congestion by dropping packets. After an ECN-capable host negotiates ECN, it signals impending congestion by marking the IP header of packets encountering the congestion instead of dropping the packets. The recipient echoes the congestion indication back to the sender, which reduces its transmission rate as if it had detected a dropped packet.

Switches support ECN for unicast queues through Weighted Random Early Detection (WRED), an active queue management (AQM) algorithm that extends Random Early Detection (RED) to define multiple thresholds for an individual queue. WRED determines congestion by comparing average queue size with queue thresholds. Average queue size depends on the previous average and current queue size:

average queue size = (old_avg * (1-2^(-weight))) + (current_queue_size * 2^(-weight))

where weight is the exponential weight factor used for averaging the queue size.

Packets are marked based on WRED as follows:

  • If average queue size is below the minimum threshold, packets are queued as in normal operation without ECN.
  • If average queue size is greater than the maximum threshold, packets are marked for congestion.
  • If average queue size is between minimum and maximum queue threshold, packets are either queued or marked. The proportion of packets that are marked increases linearly from 0% at the minimum threshold to 100% at the maximum threshold.

Treatment of packets marked as not ECN capable varies by platform.

These sections describe ECN configuration procedures:

ACL Policing

ACL policing monitors the ingress data rates for a particular class of traffic and performs the action configured when the traffic exceeds the user configured value. Therefore, it allows the user to control ingress bandwidth based on packet classification. The incoming traffic is metered and marked by the policing, and based on the metering results the actions are performed.

ACL policing uses a token bucket shaping algorithm for packet transmission. Packets are eligible for transmission when token count is positive, and when token count is negative the next packet will have to wait until the token count turns positive again. The tokens are renewed at 96ns time interval (Tc). The tokens are collected in the policer bucket up to a max burst size of 16KB, and any traffic beyond this shape rate and burst size is buffered in the shared memory. The packets are dropped if there is a memory overflow.

Note: The policer bucket is refilled at a sweeper period of 0.333 millisecond. This is applicable for all the platforms.

For example, let us assume that shaping is not enabled, and the link is at 10 Gbps, that is 1.25 bytes/nsec. In such case a each refill cycle will add tokens worth 120 bytes. For a shape rate of 500 Mbps, each refill cycle will add 6 bytes. And for 64 byte worth of tokens we need around 11 refill cycles = 1us. A 64 byte packet coming immediately after a jumbo frame will have to wait longer compared to a jumbo frame coming after 64 byte packet.

Token size depends on the interface speed, following the last example:

  • For 10 Gbps, each refill cycle will add tokens worth 120 bytes.
  • For 1 Gbps, each refill cycle will add tokens worth 12 bytes.

At lower shaping rates (less than 10 Mbps), granularity and rounding errors may alter the actual shaping rate by 20% from the specified rate, and the rounding errors are much less at higher speeds. For example, At 100 Mbps you will see 98.9Mbps configured in hardware. User can use the show qos interfaces command to verify the interface speed.

The policing uses three types of traffic metering and coloring mechanisms.

  • Single Rate Two Color Marker
  • Single Rate Three Color Marker
  • Two Rate Three Color Marker

Single Rate Two Color Marker

It meters the packet stream and marks packets based on committed burst size (bc) and excess burst size (be).

Single Rate Three Color Marker

It meters the packet stream and marks packets based on single rate committed information rate (cir), and committed burst size (bc) and excess burst size (be). The packets are marked in green if it does not exceed the set burst size, and marked in yellow if it does exceed the burst size but not the excess burst size, and marked red otherwise. The packets are marked in two color modes.

  • Color-blind Mode: In color-blind mode the incoming packet color is ignored.
  • Color-aware Mode: In color-aware mode it is assumed that incoming packet is colored by preceding entity. And, in color-aware mode, a packet never get better than it was. If the input color of the packet is green, it can be marked as green, yellow, or red. But if the input color is yellow, then it can be marked only yellow or red

Two Rate Three Color Marker

It meters the packet stream and marks its packets based on two rates, peak information rate (pir) and committed information rate (cir), and associated burst sizes (bc and be).The packet is marked red if rate exceeds ‘pir’, and yellow if it exceeds ‘cir’ but not 'pir' and marked green if rate is lower than 'cir'. The two rate mode is configured by setting four parameters pir, cir, bc, and be.

The ACL policing is supported on platforms specified in the table below.

ACL Policing Support Matrix

 

Platform Supported

ACL Policing

ACL Policing on LAG Interface

Trident

Yes

Yes

Trident II

Yes

Yes

Trident+

Yes

Yes

FM6000

Yes

Yes

Arad

Yes

Only Per-Port

Jericho

Yes

Yes

Helix

Yes

Yes

XP

Yes

Yes

Trident 3

Yes

Yes

Tomahawk

Yes

Yes

Tomahawk 2

Yes

Yes

Tofino

No

No

Configuring ACL Policing

The policer is applied to the class inside the policy map. Policy maps can contain one or more policy map classes, each with different match criteria and policers. The following is the default behavior on conditions and available policing actions:

  • Police command creates a per-interface policer. If you attach per-interface policers to multiple ingress ports, each one polices the matched traffic on each ingress port separately. Per interface statistics gathered for conformed/allowed traffic and exceeded/dropped traffic
  • If there is no policer configured within a class, all traffic is transmitted without any policing. If there are any actions configured, the configured actions are applied

    • conform-action (green): transmit (default)
    • exceed-action (yellow): drop (default)
    • violate-action (red): drop (default)
  • The policer bucket is refilled at a sweeper period of 0.333 milliseconds, and the tokens in the policer bucket are renewed at 96ns time interval (Tc). This is applicable for all the platforms.

Steps to Configure ACL Policing

These commands set the CIR, burst size, and creates a class and applies the policing to the policy map:

  1. Create a policy map.
  2. Create a class-map.
  3. Apply the policer to the policy map created.
Examples
  • These commands configure the ACL policing for a policy map.
    switch#configure terminal
    switch(config)#policy-map [type qos] policy-name
    switch(config-pmap)#class { class-name }
    switch(config-pmap-c)#[no] police cir cir [{bps|kbps|mbps}] bc 
    committed-burst-size [{bytes|kbytes|mbytes}]
  • These commands configure ACL policing in single-rate, two-color mode.
    switch(config)#class-map type qos match-any class1 
    switch(config-cmap-class1)#match ip access-group acl1
    switch(config-cmap-class1)#exit
    
    switch(config)#policy-map type quality-of-service policy1
    switch(config-pmap)#class class1 
    switch(config-pmap-c)#police cir 512000 bc 96000 
    switch(config-pmap-c)#exit 
    switch(config-pmap)#

Displaying ACL Policing Information

Examples

  • This command shows the contents of all policy maps on the switch.

    switch(config)#show policy-map 
    Service-policy p
    
     Class-map: c (match-any)
    Match: ip access-group name a
     police rate 1000 mbps burst-size 100 bytes
     Class-map: class-default (match-any)
    Service-policy p
     Class-map: c (match-any)
    Match: ip access-group name a
     police rate 1000 mbps burst-size 100 bytes
     Class-map: class-default (match-any)
  • This command shows the interface-specific police counters for Ethernet interface 1.

    switch(config)#show policy-map interface Ethernet 1 input counters
    Service-policy input: policy1
    Hardware programming status: Successful
    
    Class-map: class1 (match-any) Match: ip access-group name acl1
    Police cir 512000 bps bc 96000 bytes Conformed 4351 packets, 1857386 bytes
    Conformed 2536 packets, 3384260 bytes
    
    Class-map: class-default (match-any) matched packets: 0
  • This command shows the counters associated with the policy map called “p1.”

    switch(config)#show policy-map type qos p1 input counters
    Service-policy input: p1 Class-map: c1 (match-any)
    Match: ip access-group name a1
    Police cir 512000 bps bc 96000 bytes Interface: Ethernet1
    Conformed 4351 packets, 1857386 bytes
    Exceeded 2536 packets, 3384260 bytes
    Interface: Ethernet2
    Conformed 2351 packets, 957386 bytes
    Exceeded 1536 packets, 1384260 bytes
    Class-map: class-default (match-any)
    Matched packets : 3229
  • This command shows the QoS policy map for Ethernet interface 1.

    switch(config)#show policy-map interface Ethernet 1 input type qos
    Interface: Ethernet 1 Service-policy input: policy1
    Hardware programming status: Successful Class-map: class1 (match-any)
    Match: ip access-group name acl1
    Police cir 512000 bps bc 9000 bytes
    Class-map: class2 (match-any)
    Match: ip access-group name acl2 set dscp 2
    Class-map: class3 (match-any) Match: ip access-group name acl3
    Police cir 1280000 bps bc 9000 bytes
    Class-map: class-default (match-any)

Quality of Service (QoS) Profiles

Quality of Service (QoS) profiles are sets of QoS configuration instructions defined and applied at the interface level. A QoS profile serves the traffic better by reducing disorder in the running configuration. QoS profiles can modify all interface-level QoS configurations, and are supported on fabric, Ethernet, and port-channel interfaces. Control-plane policies cannot be applied using QoS profiles. Because configuration can be applied through QoS profiles or directly at the interface level, multiple configurations can be applied to the same interface. In such cases, QoS configurations with non-default values, whether configured through the CLI at the interface level or through a QoS profile, are given priority. In the case of multiple non-default values being configured, the interface-level CLI configuration is given priority.

Policy maps incorporating traffic resolution commands can also be applied by a QoS profile. If two policy maps are applied to the same interface (one through a QoS profile and another directly to the CLI).

Policy maps cannot be used on fabric interfaces. If a QoS profile which includes a policy map is applied to a fabric interface, a warning message will be displayed and the policy map will not be applied to the interface, but any additional supported configurations in the QoS profile will be applied. On SVIs and subinterfaces, QoS profiles are not supported, so policy maps must be applied directly through the CLI for these interfaces.

Note: For tx-queue configuration, conflicts between QoS profiles and configuration entered via the CLI are resolved at the tx-queue level and not at the tx-queue attribute level. If any non-default configuration has been entered for the tx-queue through the CLI, all tx-queue configuration included in the QoS profile is ignored.

QoS Configuration: Platform-Independent Features

Creating QoS Profiles

QoS profiles are created by using the qos profile command. This also places the switch in QoS profile configuration mode, where the QoS parameters applied to interfaces are configured. To delete a QoS profile from the running configuration, use the no form of the command.

Example

This command creates a QoS profile named “Test-Profile” and places the switch in QoS profile configuration mode for the profile.
switch(config)#qos profile Test-Profile 
switch(config-qos-profile-Test-Profile)#

Configuring QoS Profiles

The parameters that a QoS profile will apply to interfaces are configured in QoS profile configuration mode by issuing the same QoS configuration commands that are available in interface configuration mode. QoS profile configuration mode is a group change mode, and changes made in the mode are not saved until the mode is exited. To abandon all changes made while in the mode, use the abort command.

Example

These commands enter QoS profile configuration mode for a QoS profile named “Test Profile,” configure the CoS value and transmit queue, and save the changes to the profile.
switch(config)#qos profile Test-Profile
switch(config-qos-profile-Test-Profile)#qos cos 3
switch(config-qos-profile-Test-Profile)#priority-flow-control on
switch(config-qos-profile-Test-Profile)#exit

switch(config)#

Attaching Policy-Map to a QoS Profile

The qos profile command places the switch in QoS profile configuration mode. The profile applies the QoS configurations to Ethernet and Port-Channel, and even to the Fabric interfaces, if it exists. A profile specifies the policy-map and other QoS supported configurations. The policy-map is then attached to the QoS profile using service-policy command.

Profiles are created in QoS-profile configuration mode, then applied to an interface in interface configuration mode.

Examples
  • This command places the switch in QoS profile configuration mode, the policy-map is then attached to the profile using service-policy command in this mode.
    switch(config)#qos profile TP
    switch(config-qos-profile-TP)#
  • This command applies the policy-map to the QoS profile.
    switch(config-qos-profile-TP)#service-policy type qos input PM-1 

Applying a QoS profile on an Interface

The service-profile command applies a QoS profile to the configuration mode interface.

Example

This command applies the QoS profile TP to Ethernet interface 13.
switch(config)#interface ethernet 13
switch(config-if-Et13)#service-profile TP

Displaying the QoS Profile Information

The show qos profile command displays information about the QoS profiles configured and their parameters. To display the attribute of a specific profile, add the name of the profile. To display a list of configured QoS profiles and the interfaces on which they are configured, add the summary keyword.

Examples

  • This command displays the configured profiles and their configuration.

    switch#show qos profile
    qos profile p
     qos cos 1
    no priority-flow-control pause watchdog
    priority-flow-control priority 1 no-drop
    priority-flow-control priority 2 no-drop
    qos profile p2
    qos cos 3
    priority-flow-control priority 0 no-drop
    
  • This command displays the contents of a specific profile.

    switch#show qos profile p2
    qos profile p2
    qos cos 3
    priority-flow-control priority 0 no-drop
    
  • This command displays the interfaces on which each profile is applied.

    switch#show qos profile summary
    Qos Profile: p
    Configured on: Et13,7
    Fabric
    Po12
    Qos Profile: p2
    Configured on: Et56
    

QoS Configuration: Arad Platform Switches

Implementing QoS on an Arad platform switch consists of configuring port trust settings, default port settings, default traffic classes, conversion maps, and transmit queues.

CoS and DSCP Port Settings – Arad Platform Switches

 

Port Settings – Trust Mode and Traffic Class describes port trust and default port CoS and DSCP values.

Configuring Port Trust Settings

The qos trust command configures the QoS port trust mode for the configuration mode interface. Trust enabled ports use packet CoS or DSCP values to classify traffic. The port-trust default for switched ports is CoS. The port-trust default for routed ports is DSCP.

  • qos trust cos specifies CoS as the port’s port-trust mode.
  • qos trust dscp specifies DSCP as the port’s port-trust mode.
  • no qos trust specifies untrusted as the port’s port-trust mode.

The show qos interfaces trust command displays the trust mode of specified interfaces.

Example
  • These commands configure and display the following trust modes:

    • Ethernet 3/5/1: dscp
    • Ethernet 3/5/2: untrusted
    • Ethernet 3/5/3: cos
    • Ethernet 3/5/4: default as a switched port
    • Ethernet 3/6/1: default as a routed port

      switch(config)#interface ethernet 3/5/1
      switch(config-if-Et3/5/1)#qos trust dscp
      switch(config-if-Et3/5/1)#interface ethernet 3/5/2
      switch(config-if-Et3/5/2)#no qos trust
      switch(config-if-Et3/5/2)#interface ethernet 3/5/3
      switch(config-if-Et3/5/3)#qos trust cos
      switch(config-if-Et3/5/3)#interface ethernet 3/5/4
      switch(config-if-Et3/5/4)#switchport
      switch(config-if-Et3/5/4)#default qos trust
      switch(config-if-Et3/5/4)#interface ethernet 3/6/1
      switch(config-if-Et3/6/1)#no switchport
      switch(config-if-Et3/6/1)#default qos trust
      switch(config-if-Et3/6/1)#show qos interface ethernet 3/5/1 - 3/6/1 trust
      Port Trust Mode
       Operational Configured
      ---------------------------------------------------------------
      Ethernet3/5/1DSCPDSCP
      Ethernet3/5/2UNTRUSTED UNTRUSTED
      Ethernet3/5/3COS COS
      Ethernet3/5/4COS DEFAULT
      Ethernet3/6/1DSCPDEFAULT
      
      switch(config-if-Et3/6/1)#
      

Configuring Default Port Settings

Default CoS and DSCP values are assigned to each Ethernet and port channel interface. These commands specify the configuration mode interface commands specify the port’s default CoS and DSCP values.

  • qos cos configures a port’s default CoS value.
  • qos dscp configures a port’s default DSCP value.
Example
These commands configure default CoS (4) and DSCP (44) values on Ethernet 
interface 3/6/2.
switch(config)#interface ethernet 3/6/2
switch(config-if-Et3/6/2)#qos cos 4
switch(config-if-Et3/6/2)#qos dscp 44
switch(config-if-Et3/6/2)#show active
interface Ethernet3/6/2
 qos cos 4
 qos dscp 44
switch(config-if-Et3/6/2)#show qos interfaces ethernet 3/6/2
Ethernet3/6/2:
 Trust Mode: COS
 Default COS: 4
 Default DSCP: 44

switch(config-if-Et3/6/2)#

Traffic Class Derivations – Arad Platform Switches

Traffic Classesdescribes traffic classes.

Traffic Class Derivation Source

The following table displays the source for deriving a data stream’s traffic class.

Table 1. Traffic Class Derivation Source: Arad Platform Switches
  Untrusted CoS Trusted DSCP Trusted
Untagged Non-IP Default CoS (port) Default CoS (port) Default DSCP (port)
Untagged IP Default CoS (port) Default CoS (port) DSCP (packet)
Tagged Non-IP Default CoS (port) CoS (packet) Default DSCP (port)
Tagged IP Default CoS (port) CoS (packet) DSCP (packet)

CoS and DSCP Port Settings – Arad Platform Switchesdescribes the default CoS and DSCP settings for each port.

Mapping CoS to Traffic Class

The qos map cos command assigns a traffic class to a list of CoS values. Multiple commands create a complete CoS to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s CoS field or the chip upon which it is received.

Example

This command assigns the traffic class of 5 to the classes of service 1, 3, 5, and 7.
switch(config)#qos map cos 1 3 5 7 to traffic-class 5
switch(config)#show qos maps
 Number of Traffic Classes supported: 8


 Cos-tc map:
 cos:01234567
 ----------------------------
 tc: 15254565


switch(config)#

The following table displays the default CoS to Traffic Class map on Arad platform switches.

Table 2. Default CoS to Traffic Class Map: Arad Platform Switches
Inbound CoS Untagged 0 1 2 3 4 5 6 7
Traffic Class Derived: use default CoS as inbound 1 0 2 3 4 5 6 7

Mapping DSCP to Traffic Class

The qos map dscp command assigns a traffic class to a set of DSCP values. Multiple commands create a complete DSCP to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s DSCP field or the chip upon which it is received.

Example

This command assigns the traffic class of 0 to DSCP values of 12, 24, 41, and 44-47.
switch(config)#qos map dscp 12 24 41 44 45 46 47 to traffic-class 0
switch(config)#show qos maps
 Number of Traffic Classes supported: 8


 Dscp-tc map:
 d1 :d2 0123456789
 --------------------------------------
0 : 1111111100
1 : 0000002222
2 : 2222033333
3 : 3344444444
4 : 5055000066
5 : 6666667777
6 : 7777


switch(config)#

The following table displays the default DSCP to traffic class map on Arad platform switches.

Table 3. Default DSCP to Traffic Class Map: Arad Platform Switches
Inbound DSCP 0-7 8-15 16-23 24-31 32-39 40-47 48-55 56-63
Traffic Class 1 0 2 3 4 5 6 7

CoS Rewrite – Arad Platform Switches

Rewriting CoS and DSCP describes the CoS rewrite functions.

Traffic Class to CoS Rewrite Map

The CoS rewrite value is configurable and based on a data stream’s traffic class, as specified by the traffic class-CoS rewrite map. The qos map traffic-class to coscommand assigns a CoS rewrite value to a list of traffic classes. Multiple commands create the complete traffic class–CoS rewrite map.

Example

This command assigns the CoS of two to traffic classes 1, 3, and 5.
switch(config)#qos map traffic-class 1 3 5 to cos 2
switch(config)#show qos map
 Number of Traffic Classes supported: 8


 Tc-cos map:
 tc: 01234567
 ----------------------------
 cos:12224267


switch(config)#

The following table displays the default Traffic Class to CoS rewrite value map on Arad platform switches.

Table 4. Default Traffic Class to CoS Rewrite Value Map: Arad Platform Switches

Traffic Class

0

1

2

3

4

5

6

7

CoS Rewrite Value

1

0

2

3

4

5

6

7

Traffic Class to DSCP Rewrite Map

DSCP rewrite is always disabled on Arad platform switches.

Transmit Queues and Port Shaping – Arad Platform Switches

Transmit Queues and Port Shaping describes transmit queues and port shaping.

Arad platform switches provide 16 physical queues for each egress port: eight unicast and eight multicast queues. Data is scheduled to the physical queues based on transmit queue assignments.

Multicast queue capacity that remains after multicast traffic is serviced is available for unicast traffic of a corresponding priority. Similarly, unicast queue capacity that remains after unicast traffic is serviced is available for overflow multicast traffic. Under conditions of unicast and multicast congestion, egress traffic is evenly split between unicast and multicast traffic.

A data stream’s traffic class determines the transmit queue it uses. The switch defines a single traffic class–transmit queue map for unicast and multicast traffic on all Ethernet and port channel interfaces. The show qos maps command displays the traffic class–transmit queue map.

The following tabledisplays the default traffic class to transmit queue map on Arad platform switches.

Table 5. Default Traffic Class to Transmit Queue Map: Arad Platform Switches
Traffic Class 0 1 2 3 4 5 6 7
Transmit Queue 0 1 2 3 4 5 6 7

Transmit queue parameters are configured in tx-queue configuration command mode, which is entered from interface-ethernet configuration mode.

Mapping Traffic Classes to a Transmit Queue

The qos map traffic-class to tx-queue command assigns traffic classes to a transmit queue. Multiple commands complete the traffic class-transmit queue map. Traffic class 7 and transmit queue 7 are always mapped to each other. This association is not editable.

Example

These commands assign traffic classes of 1, 3, and 5 to transmit queue 1, traffic classes 2, 4, and 6 to transmit queue 2, and traffic class 0 to transmit queue 0, then display the resultant map.
switch(config)#qos map traffic-class 1 3 5 to tx-queue 1
switch(config)#qos map traffic-class 2 4 6 to tx-queue 2
switch(config)#qos map traffic-class 0 to tx-queue 0
switch(config)#show qos maps
 Number of Traffic Classes supported: 8
 Number of Transmit Queues supported: 8


 Tc - tx-queue map:
 tc:01234567
 ---------------------------------
 tx-queue:01212127

switch(config)#

Entering Tx-Queue Configuration Mode

The tx-queue (Arad/Jericho) command places the switch in tx-queue configuration mode to configure a transmit queue on the configuration mode interface. Tx-queue 7 is not configurable. The show qos interfaces displays the transmit queue configuration for a specified port.

Example

This command enters Tx-queue configuration mode for transmit queue 4 of Ethernet interface 3/3/3.
switch(config)#interface ethernet 3/3/3
switch(config-if-Et3/3/3)#tx-queue 4
switch(config-if-Et3/3/3-txq-4)#

Configuring the Shape Rate – Port and Transmit Queues

A port’s shape rate specifies its maximum outbound traffic bandwidth. A transmit queue’s shape rate specifies the queue’s maximum outbound bandwidth. Shape rate commands specify data rates in kbps.

Examples
  • This command configures a port shape rate of 5 Gbps on Ethernet interface 3/5/1.
    switch(config)#interface ethernet 3/5/1
    switch(config-if-Et3/5/1)#shape rate 5000000
    switch(config-if-Et3/5/1)#show qos interfaces ethernet 3/5/1 
    Ethernet3/5/1:
    
     Port shaping rate: 5000012 / 5000000 kbps
    
    TxBandwidth Shape RatePriorityECN
     Queue(percent)(units)
     -----------------------------------------------------
     7- / -- / -(-)SP / SPD
    
    switch(config-if-Et3/5/1)#
  • These commands configure a shape rate of 1 Gbps on transmit queues 3 and 4 of Ethernet interface 3/4/1.
    switch(config)#interface ethernet 3/4/1
    switch(config-if-Et3/4/1)#tx-queue 4
    switch(config-if-Et3/4/1-txq-4)#shape rate 1000000 kbps
    switch(config-if-Et3/4/1-txq-4)#tx-queue 3
    switch(config-if-Et3/4/1-txq-3)#shape rate 1000000 kbps
    switch(config-if-Et3/4/1-txq-3)#show qos interface ethernet 3/4/1
    Ethernet3/4/1:
    
     Port shaping rate: disabled
    
    TxBandwidth Shape RatePriorityECN
     Queue(percent)(units)
     -----------------------------------------------------
     7- / -- / -(-)SP / SPD
     6- / -- / -(-)SP / SPD
     5- / -- / -(-)SP / SPD
     4- / -999 / 1000 ( Mbps ) SP / SPD
     3- / -999 / 1000 ( Mbps ) SP / SPD
     2- / -- / -(-)SP / SPD
     1- / -- / -(-)SP / SPD
     0- / -- / -(-)SP / SPD
    
    switch(config-if-Et3/4/1-txq-3)#

Configuring Queue Priority

The priority (Arad/Jericho) command configures a transmit queue’s priority type:

  • The priority strict command configures the queue as a strict priority queue.
  • The no priority command configures the queue as a round robin queue.

A queue’s configuration as round robin also applies to all lower priority queues regardless of other configuration statements.

The bandwidth percent (Arad/Jericho) command configures a round robin queue’s bandwidth share. The cumulative operational bandwidth of all round robin queues is always less than or equal to 100%. If the cumulative configured bandwidth is greater than 100%, each port’s operational bandwidth is its configured bandwidth divided by the cumulative configured bandwidth.

Examples
  • These commands configure queues 0 through 3 (Ethernet interface 3/5/1) as round robin, then allocate bandwidth for three queues at 30% and one queue at 10%.
    The no priority statement for queue 3 also configures queues 0, 1, and 2 as round robin queues. Removing this statement reverts the other queues to strict priority type unless running-config contains a no priority statement for one of these queues.
    switch(config)#interface ethernet 3/5/1
    switch(config-if-Et3/5/1)#tx-queue 3
    switch(config-if-Et3/5/1-txq-3)#no priority
    switch(config-if-Et3/5/1-txq-3)#bandwidth percent 10
    switch(config-if-Et3/5/1-txq-3)#tx-queue 2
    switch(config-if-Et3/5/1-txq-2)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-2)#tx-queue 1
    switch(config-if-Et3/5/1-txq-1)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-1)#tx-queue 0
    switch(config-if-Et3/5/1-txq-0)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-0)#show qos interfaces ethernet 3/5/1
    Ethernet3/5/1:
    
    TxBandwidth Shape RatePriorityECN
     Queue(percent)(units)
     -----------------------------------------------------
     7- / -- / -(-)SP / SPD
     6- / -- / -(-)SP / SPD
     5- / -- / -(-)SP / SPD
     4- / -- / -(-)SP / SPD
     3 10 / 10 - / -(-)RR / RRD
     2 30 / 30 - / -(-)RR / SPD
     1 30 / 30 - / -(-)RR / SPD
     0 30 / 30 - / -(-)RR / SPD
    
    switch(config-if-Et3/5/1-txq-0)#
    
  • Changing the bandwidth percentage for queue 3 to 30 changes the operational bandwidth of each queue to its configured bandwidth divided by 120% (10%+20%+30%+60%).
    switch(config-if-Et3/5/1-txq-0)#tx-queue 3
    switch(config-if-Et3/5/1-txq-3)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-3)#show qos interfaces ethernet 3/5/1
    Ethernet3/5/1:
    
     Port shaping rate: disabled
    
    TxBandwidth Shape RatePriorityECN
     Queue(percent)(units)
     -----------------------------------------------------
     7- / -- / -(-)SP / SPD
     6- / -- / -(-)SP / SPD
     5- / -- / -(-)SP / SPD
     4- / -- / -(-)SP / SPD
     3 24 / 30 - / -(-)RR / RRD
     2 24 / 30 - / -(-)RR / SPD
     1 24 / 30 - / -(-)RR / SPD
     0 24 / 30 - / -(-)RR / SPD
    
    Note: Values are displayed as Operational/Configured
    
    switch(config-if-Et3/5/1-txq-3)#

ECN Configuration – Arad Platform Switches

Explicit Congestion Notification (ECN)describes Explicit Congestion Notification (ECN).

ECN is independently configurable on all egress queues of each Ethernet interface. ECN settings for Port-Channels are applied on each of the channel’s member Ethernet interfaces. Average queue length is tracked for transmit queues. When it reaches maximum threshold, all subsequent packets are marked.

Although the switch does not limit the number of queues that can be configured for ECN, hardware table limitations restrict the number of queues that can simultaneously implement ECN.

The random-detect ecn (Arad/Jericho) command enables ECN marking for the configuration mode unicast transmit queue and specifies threshold queue sizes.

Example

These commands enable ECN marking of unicast packets from unicast transmit queue 4 of Ethernet interface 3/5/1, setting thresholds at 128 kbytes and 1280 kbytes.
switch(config)#interface ethernet 3/5/1
switch(config-if-Et3/5/1)#tx-queue 4
switch(config-if-Et3/5/1-txq-4)#random-detect ecn minimum-threshold 128 kbytes 
maximum-threshold 1280 kbyte
switch(config-if-Et3/5/1-txq-4)#show active
interface Ethernet3/5/1
 tx-queue 4
random-detect ecn minimum-threshold 128 kbytes maximum-threshold 1280 kbytes
switch(config-if-Et3/5/1-txq-4)#

ACL Policing – Arad Platform Switches

ACL Policing describes ACL policing.

Implementing ACL policing consists of configuring the following:

  • policy-map settings
  • class-name
  • committed information rate (CIR) the data speed committed to any given circuit regardless of the number of users
  • burst size the maximum burst size in bytes the network commits to moving under normal conditions

The default unit for the metering rate CIR is bits per second; the default unit for the burst size is bytes.

The policer is applied to the class inside the policy map. Policy maps can contain one or more policy map classes, each with different match criteria and policer.

Default behavior and available policing actions are as follows:

  • Policy map can be applied on multiple interfaces. Interfaces on the same chip will share the policer. (Applicable for Arad only.)
  • If there is no policer configured within a class, all traffic is transmitted without any policing.
  • If there are any actions configured, the configured actions are applied:
    • Conform-action (green): transmit (default)
    • Violate-action (red): drop (default)

Example

These commands configure ACL policing in single-rate, two-color mode.

switch(config)#class-map type qos match-any class1
switch(config-cmap-class1)#match ip access-group acl1
switch(config-cmap-class1)#exit
switch(config)#policy-map type quality-of-service policy1
switch(config-policy1)#class class1
switch(config-policy1-class1)#police cir 512000 bc 96000
switch(config-policy1-class1)#exit
switch(config-policy1)#exit
switch(config)#

Displaying ACL Policing Information

Examples

  • This command shows the contents of all policy maps on the switch.

    switch(config)#show policy-map
    Service-policy policy1
    
    Class-map: class1 (match-any)
    Match: ip access-group name acl1
    Police cir 512000 bps bc 96000 bytes
    
    Class-map: class-default (match-any)
    switch(config)#
    
  • This command shows the interface-specific police counters for Ethernet interface 1.

    switch(config)#show policy-map interface Ethernet 1 input counters
    Service-policy input: policy1
    Hardware programming status: Successful
    
    Class-map: class1 (match-any) Match: ip access-group name acl1
    Police cir 512000 bps bc 96000 bytes Conformed 4351 packets, 1857386 bytes
    Conformed 2536 packets, 3384260 bytes
    
    Class-map: class-default (match-any) matched packets: 0
    
    switch(config)#
    
  • This command shows the counters associated with the policy map called “p1.”

    switch(config)#show policy-map type qos p1 input counters
    Service-policy input: p1 Class-map: c1 (match-any)
    Match: ip access-group name a1
    Police cir 512000 bps bc 96000 bytes Interface: Ethernet1
    Conformed 4351 packets, 1857386 bytes
    Exceeded 2536 packets, 3384260 bytes
    Interface: Ethernet2
    Conformed 2351 packets, 957386 bytes
    Exceeded 1536 packets, 1384260 bytes
    Class-map: class-default (match-any)
    Matched packets : 3229
    
    switch(config)#
    
  • This command shows the QoS policy map for Ethernet interface 1.

    switch(config)#show policy-map interface Ethernet 1 input type qos
    Interface: Ethernet 1 Service-policy input: policy1
    Hardware programming status: Successful Class-map: class1 (match-any)
    Match: ip access-group name acl1
    Police cir 512000 bps bc 9000 bytes
    Class-map: class2 (match-any)
    Match: ip access-group name acl2 set dscp 2
    Class-map: class3 (match-any) Match: ip access-group name acl3
    Police cir 1280000 bps bc 9000 bytes
    Class-map: class-default (match-any)
    
    switch(config)#
    

QoS Configuration: Jericho Platform Switches

Implementing QoS on an Jericho platform switch consists of configuring port trust settings, default port settings, default traffic classes, conversion maps, and transmit queues.

CoS and DSCP Port Settings – Jericho Platform Switches

Port Settings – Trust Mode and Traffic Class describes port trust and default port CoS and DSCP values.

Configuring Port Trust Settings

The qos trust command configures the QoS port trust mode for the configuration mode interface. Trust enabled ports use packet CoS or DSCP values to classify traffic. The port-trust default for switched ports is CoS. The port-trust default for routed ports is DSCP.

  • qos trust cos specifies CoS as the port’s port-trust mode.
  • qos trust dscp specifies DSCP as the port’s port-trust mode.
  • no qos trust specifies untrusted as the port’s port-trust mode.

The show qos interfaces trust command displays the trust mode of specified interfaces.

Example
  • These commands configure and display the following trust modes:

    • Ethernet 3/5/1: dscp
    • Ethernet 3/5/2: untrusted
    • Ethernet 3/5/3: cos
    • Ethernet 3/5/4: default as a switched port
    • Ethernet 3/6/1: default as a routed port

      switch(config)#interface ethernet 3/5/1
      switch(config-if-Et3/5/1)#qos trust dscp
      switch(config-if-Et3/5/1)#interface ethernet 3/5/2
      switch(config-if-Et3/5/2)#no qos trust
      switch(config-if-Et3/5/2)#interface ethernet 3/5/3
      switch(config-if-Et3/5/3)#qos trust cos
      switch(config-if-Et3/5/3)#interface ethernet 3/5/4
      switch(config-if-Et3/5/4)#switchport
      switch(config-if-Et3/5/4)#default qos trust
      switch(config-if-Et3/5/4)#interface ethernet 3/6/1
      switch(config-if-Et3/6/1)#no switchport
      switch(config-if-Et3/6/1)#default qos trust
      switch(config-if-Et3/6/1)#show qos interface ethernet 3/5/1 - 3/6/1 trust
      Port Trust Mode
       Operational Configured
      ---------------------------------------------------------------
      Ethernet3/5/1DSCPDSCP
      Ethernet3/5/2UNTRUSTED UNTRUSTED
      Ethernet3/5/3COS COS
      Ethernet3/5/4COS DEFAULT
      Ethernet3/6/1DSCPDEFAULT
      
      switch(config-if-Et3/6/1)#

Configuring Default Port Settings

Default CoS and DSCP values are assigned to each Ethernet and port channel interface. These commands specify the configuration mode interface commands specify the port’s default CoS and DSCP values.

  • qos cos configures a port’s default CoS value.
  • qos dscp configures a port’s default DSCP value.
Example
These commands configure default CoS (4) and DSCP (44) values on Ethernet 
interface 3/6/2.
switch(config)#interface ethernet 3/6/2
switch(config-if-Et3/6/2)#qos cos 4
switch(config-if-Et3/6/2)#qos dscp 44
switch(config-if-Et3/6/2)#show active
interface Ethernet3/6/2
 qos cos 4
 qos dscp 44
switch(config-if-Et3/6/2)#show qos interfaces ethernet 3/6/2
Ethernet3/6/2:
 Trust Mode: COS
 Default COS: 4
 Default DSCP: 44

switch(config-if-Et3/6/2)#

Traffic Class Derivations – Jericho Platform Switches

Traffic Classes describes traffic classes.

Traffic Class Derivation Source

The following table displays the source for deriving a data stream’s traffic class on Jericho platform switches.

Table 6. Traffic Class Derivation Source: Jericho Platform Switches
  Untrusted CoS Trusted DSCP Trusted
Untagged Non-IP Default CoS (port) Default CoS (port) Default DSCP (port)
Untagged IP Default CoS (port) Default CoS (port) DSCP (packet)
Tagged Non-IP Default CoS (port) CoS (packet) Default DSCP (port)
Tagged IP Default CoS (port) CoS (packet) DSCP (packet)

CoS and DSCP Port Settings – Arad Platform Switches describes the default CoS and DSCP settings for each port.

Mapping CoS to Traffic Class

The qos map cos command assigns a traffic class to a list of CoS values. Multiple commands create a complete CoS to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s CoS field or the chip upon which it is received.

Example

This command assigns the traffic class of 5 to the classes of service 1, 3, 5, and 7.
switch(config)#qos map cos 1 3 5 7 to traffic-class 5
switch(config)#show qos maps
 Number of Traffic Classes supported: 8

 Cos-tc map:
 cos:01234567
 ----------------------------
 tc: 15254565

switch(config)#

The following table displays the default CoS to Traffic Class map on Jericho platform switches.

Table 7. Default CoS to Traffic Class Map: Jericho Platform Switches
Inbound CoS Untagged 0 1 2 3 4 5 6 7
Traffic Class Derived: use default CoS as inbound 1 0 2 3 4 5 6 7

Mapping DSCP to Traffic Class

The qos map dscp command assigns a traffic class to a set of DSCP values. Multiple commands create a complete DSCP to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s DSCP field or the chip upon which it is received.

Example

This command assigns the traffic class of 0 to DSCP values of 12, 24, 41, and 44-47.
switch(config)#qos map dscp 12 24 41 44 45 46 47 to traffic-class 0
switch(config)#show qos maps
 Number of Traffic Classes supported: 8


 Dscp-tc map:
 d1 :d2 0123456789
 --------------------------------------
0 : 1111111100
1 : 0000002222
2 : 2222033333
3 : 3344444444
4 : 5055000066
5 : 6666667777
6 : 7777


switch(config)#

The following table displays the default DSCP to traffic class map on Jericho platform switches.

Table 8. Default DSCP to Traffic Class Map: Jericho Platform Switches
Inbound DSCP 0-7 8-15 16-23 24-31 32-39 40-47 48-55 56-63
Traffic Class 1 0 2 3 4 5 6 7

CoS Rewrite – Jericho Platform Switches

Rewriting CoS and DSCPdescribes the CoS rewrite functions.

Traffic Class to CoS Rewrite Map

The CoS rewrite value is configurable and based on a data stream’s traffic class, as specified by the traffic class-CoS rewrite map. The qos map traffic-class to cos command assigns a CoS rewrite value to a list of traffic classes. Multiple commands create the complete traffic class–CoS rewrite map.

Example

This command assigns the CoS of two to traffic classes 1, 3, and 5.
switch(config)#qos map traffic-class 1 3 5 to cos 2
switch(config)#show qos map
 Number of Traffic Classes supported: 8


 Tc-cos map:
 tc: 01234567
 ----------------------------
 cos:12224267


switch(config)#

The following table displays the default Traffic Class to CoS rewrite value map on Jericho platform switches.

Table 9. Default Traffic Class to CoS Rewrite Value Map: Jericho Platform Switches

Traffic Class

0

1

2

3

4

5

6

7

CoS Rewrite Value

1

0

2

3

4

5

6

7

Traffic Class to DSCP Rewrite Map

DSCP rewrite is always disabled on Jericho platform switches.

Transmit Queues and Port Shaping – Jericho Platform Switches

Transmit Queues and Port Shapingdescribes transmit queues and port shaping.

Jericho platform switches provide 16 physical queues for each egress port: eight unicast and eight multicast queues. Data is scheduled to the physical queues based on transmit queue assignments.

Multicast queue capacity that remains after multicast traffic is serviced is available for unicast traffic of a corresponding priority. Similarly, unicast queue capacity that remains after unicast traffic is serviced is available for overflow multicast traffic. Under conditions of unicast and multicast congestion, egress traffic is evenly split between unicast and multicast traffic.

A data stream’s traffic class determines the transmit queue it uses. The switch defines a single traffic class–transmit queue map for unicast and multicast traffic on all Ethernet and port channel interfaces. The show qos maps command displays the traffic class–transmit queue map.

The following table displays the default traffic class to transmit queue map on Jericho platform switches.

Table 10. Default Traffic Class to Transmit Queue Map: Jericho Platform Switches
Traffic Class 0 1 2 3 4 5 6 7
Transmit Queue 0 1 2 3 4 5 6 7

Transmit queue parameters are configured in tx-queue configuration command mode, which is entered from interface-ethernet configuration mode.

Mapping Traffic Classes to a Transmit Queue

The qos map traffic-class to tx-queue command assigns traffic classes to a transmit queue. Multiple commands complete the traffic class-transmit queue map. Traffic class 7 and transmit queue 7 are always mapped to each other. This association is not editable.

Example

These commands assign traffic classes of 1, 3, and 5 to transmit queue 1, traffic classes 2, 4, and 6 to transmit queue 2, and traffic class 0 to transmit queue 0, then display the resultant map.
switch(config)#qos map traffic-class 1 3 5 to tx-queue 1
switch(config)#qos map traffic-class 2 4 6 to tx-queue 2
switch(config)#qos map traffic-class 0 to tx-queue 0
switch(config)#show qos maps
 Number of Traffic Classes supported: 8
 Number of Transmit Queues supported: 8


 Tc - tx-queue map:
 tc:01234567
 ---------------------------------
 tx-queue:01212127

switch(config)#

Entering Tx-Queue Configuration Mode

The tx-queue (Arad/Jericho) command places the switch in tx-queue configuration mode to configure a transmit queue on the configuration mode interface. Tx-queue 7 is not configurable. The show qos interfaces displays the transmit queue configuration for a specified port.

Example

This command enters Tx-queue configuration mode for transmit queue 4 of Ethernet interface 3/3/3.
switch(config)#interface ethernet 3/3/3
switch(config-if-Et3/3/3)#tx-queue 4
switch(config-if-Et3/3/3-txq-4)#

Configuring the Shape Rate – Port and Transmit Queues

A port’s shape rate specifies its maximum outbound traffic bandwidth. A transmit queue’s shape rate specifies the queue’s maximum outbound bandwidth. Shape rate commands specify data rates in kbps.

Examples
  • This command configures a port shape rate of 5 Gbps on Ethernet interface 3/5/1.
    switch(config)#interface ethernet 3/5/1
    switch(config-if-Et3/5/1)#shape rate 5000000
    switch(config-if-Et3/5/1)#show qos interfaces ethernet 3/5/1
    Ethernet3/5/1:
    
     Port shaping rate: 5000012 / 5000000 kbps
    
    TxBandwidth Shape RatePriorityECN
     Queue(percent)(units)
     -----------------------------------------------------
     7- / -- / -(-)SP / SPD
    
    switch(config-if-Et3/5/1)#
    
  • These commands configure a shape rate of 1 Gbps on transmit queues 3 and 4 of Ethernet interface 3/4/1.
    switch(config)#interface ethernet 3/4/1
    switch(config-if-Et3/4/1)#tx-queue 4
    switch(config-if-Et3/4/1-txq-4)#shape rate 1000000 kbps
    switch(config-if-Et3/4/1-txq-4)#tx-queue 3
    switch(config-if-Et3/4/1-txq-3)#shape rate 1000000 kbps
    switch(config-if-Et3/4/1-txq-3)#show qos interface ethernet 3/4/1
    Ethernet3/4/1:
    
     Port shaping rate: disabled
    
    TxBandwidth Shape RatePriorityECN
     Queue(percent)(units)
     -----------------------------------------------------
     7- / -- / -(-)SP / SPD
     6- / -- / -(-)SP / SPD
     5- / -- / -(-)SP / SPD
     4- / -999 / 1000 ( Mbps ) SP / SPD
     3- / -999 / 1000 ( Mbps ) SP / SPD
     2- / -- / -(-)SP / SPD
     1- / -- / -(-)SP / SPD
     0- / -- / -(-)SP / SPD
    
    switch(config-if-Et3/4/1-txq-3)#
    

Configuring Queue Priority

The priority (Arad/Jericho) command configures a transmit queue’s priority type:

  • The priority strict command configures the queue as a strict priority queue.
  • The no priority command configures the queue as a round robin queue.

A queue’s configuration as round robin also applies to all lower priority queues regardless of other configuration statements.

The bandwidth percent (Arad/Jericho) command configures a round robin queue’s bandwidth share. The cumulative operational bandwidth of all round robin queues is always less than or equal to 100%. If the cumulative configured bandwidth is greater than 100%, each port’s operational bandwidth is its configured bandwidth divided by the cumulative configured bandwidth.

Examples
  • These commands configure queues 0 through 3 (Ethernet interface 3/5/1) as round robin, then allocate bandwidth for three queues at 30% and one queue at 10%.
    The no priority statement for queue 3 also configures queues 0, 1, and 2 as round robin queues. Removing this statement reverts the other queues to strict priority type unless running-config contains a no priority statement for one of these queues.
    switch(config)#interface ethernet 3/5/1
    switch(config-if-Et3/5/1)#tx-queue 3
    switch(config-if-Et3/5/1-txq-3)#no priority
    switch(config-if-Et3/5/1-txq-3)#bandwidth percent 10
    switch(config-if-Et3/5/1-txq-3)#tx-queue 2
    switch(config-if-Et3/5/1-txq-2)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-2)#tx-queue 1
    switch(config-if-Et3/5/1-txq-1)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-1)#tx-queue 0
    switch(config-if-Et3/5/1-txq-0)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-0)#show qos interfaces ethernet 3/5/1
    Ethernet3/5/1:
    
    TxBandwidth Shape RatePriorityECN
     Queue(percent)(units)
     -----------------------------------------------------
     7- / -- / -(-)SP / SPD
     6- / -- / -(-)SP / SPD
     5- / -- / -(-)SP / SPD
     4- / -- / -(-)SP / SPD
     3 10 / 10 - / -(-)RR / RRD
     2 30 / 30 - / -(-)RR / SPD
     1 30 / 30 - / -(-)RR / SPD
     0 30 / 30 - / -(-)RR / SPD
    
    switch(config-if-Et3/5/1-txq-0)#
    
  • Changing the bandwidth percentage for queue 3 to 30 changes the operational bandwidth of each queue to its configured bandwidth divided by 120% (10%+20%+30%+60%).
    switch(config-if-Et3/5/1-txq-0)#tx-queue 3
    switch(config-if-Et3/5/1-txq-3)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-3)#show qos interfaces ethernet 3/5/1
    Ethernet3/5/1:
    
     Port shaping rate: disabled
    
    TxBandwidth Shape RatePriorityECN
     Queue(percent)(units)
     -----------------------------------------------------
     7- / -- / -(-)SP / SPD
     6- / -- / -(-)SP / SPD
     5- / -- / -(-)SP / SPD
     4- / -- / -(-)SP / SPD
     3 24 / 30 - / -(-)RR / RRD
     2 24 / 30 - / -(-)RR / SPD
     1 24 / 30 - / -(-)RR / SPD
     0 24 / 30 - / -(-)RR / SPD
    
    Note: Values are displayed as Operational/Configured
    
    switch(config-if-Et3/5/1-txq-3)#
    

ACL Policing – Jericho Platform Switches

ACL Policingdescribes ACL policing.

Implementing ACL policing consists of configuring the following:

  • policy-map settings
  • class-name
  • committed information rate (CIR)     the data speed committed to any given circuit regardless of the number of users
  • burst size     the maximum burst size in bytes the network commits to moving under normal conditions

The default unit for the metering rate CIR is bits per second; the default unit for the burst size is bytes.

The policer is applied to the class inside the policy map. Policy maps can contain one or more policy map classes, each with different match criteria and policer.

Default behavior and available policing actions are as follows:

  • Policy map can be applied on multiple interfaces. Interfaces on the same chip will share the policer. (Applicable for Arad and Jericho only.)
  • If there is no policer configured within a class, all traffic is transmitted without any policing.
  • If there are any actions configured, the configured actions are applied:
    • Conform-action (green): transmit (default)
    • Violate-action (red): drop (default)

Example

These commands configure ACL policing in single-rate, two-color mode.

switch(config)#class-map type qos match-any class1
switch(config-cmap-class1)#match ip access-group acl1
switch(config-cmap-class1)#exit
switch(config)#policy-map type quality-of-service policy1
switch(config-policy1)#class class1
switch(config-policy1-class1)#police cir 512000 bc 96000
switch(config-policy1-class1)#exit
switch(config-policy1)#exit
switch(config)#

Displaying ACL Policing Information

Examples

  • This command shows the contents of all policy maps on the switch.

    switch(config)#show policy-map
    Service-policy policy1
    
    Class-map: class1 (match-any)
    Match: ip access-group name acl1
    Police cir 512000 bps bc 96000 bytes
    
    Class-map: class-default (match-any)
    switch(config)#
  • This command shows the interface-specific police counters for Ethernet interface 1.

    switch(config)#show policy-map interface Ethernet 1 input counters
    Service-policy input: policy1
    Hardware programming status: Successful
    
    Class-map: class1 (match-any) Match: ip access-group name acl1
    Police cir 512000 bps bc 96000 bytes Conformed 4351 packets, 1857386 bytes
    Conformed 2536 packets, 3384260 bytes
    
    Class-map: class-default (match-any) matched packets: 0
    
    switch(config)#
  • This command shows the counters associated with the policy map called “p1.”

    switch(config)#show policy-map type qos p1 input counters
    Service-policy input: p1 Class-map: c1 (match-any)
    Match: ip access-group name a1
    Police cir 512000 bps bc 96000 bytes Interface: Ethernet1
    Conformed 4351 packets, 1857386 bytes
    Exceeded 2536 packets, 3384260 bytes
    Interface: Ethernet2
    Conformed 2351 packets, 957386 bytes
    Exceeded 1536 packets, 1384260 bytes
    Class-map: class-default (match-any)
    Matched packets : 3229
    
    switch(config)#
  • This command shows the QoS policy map for Ethernet interface 1.

    switch(config)#show policy-map interface Ethernet 1 input type qos
    Interface: Ethernet 1 Service-policy input: policy1
    Hardware programming status: Successful Class-map: class1 (match-any)
    Match: ip access-group name acl1
    Police cir 512000 bps bc 9000 bytes
    Class-map: class2 (match-any)
    Match: ip access-group name acl2 set dscp 2
    Class-map: class3 (match-any) Match: ip access-group name acl3
    Police cir 1280000 bps bc 9000 bytes
    Class-map: class-default (match-any)
    
    switch(config)#

QoS Configuration: FM6000 Platform Switches

Implementing QoS on an FM6000 platform switch consists of configuring port trust settings, default port settings, default traffic classes, conversion maps, and transmit queues.

CoS and DSCP Port Settings – FM6000 Platform Switches

Port Settings – Trust Mode and Traffic Class describes port trust and default port CoS and DSCP values.

Configuring Port Trust Settings

The qos trust command configures the QoS port trust mode for the configuration mode interface. Trust enabled ports use packet CoS or DSCP values to classify traffic. The port-trust default for switched ports is cos. The port-trust default for routed ports is dscp.

  • qos trust cos specifies cos as the port’s port-trust mode.
  • qos trust dscp specifies dscp as the port’s port-trust mode.
  • no qos trust specifies untrusted as the port’s port-trust mode.

The show qos interfaces trust command displays the trust mode of specified interfaces.

Example
  • These commands configure and display the following trust modes:

    • Ethernet 15: dscp
    • Ethernet 16: untrusted
    • Ethernet 17: cos
    • Ethernet 18: default as a switched port
    • Ethernet 19: default as a routed port

      switch(config)#interface ethernet 15
      switch(config-if-Et15)#qos trust dscp
      switch(config-if-Et15)#interface ethernet 16
      switch(config-if-Et16)#no qos trust
      switch(config-if-Et16)#interface ethernet 17
      switch(config-if-Et17)#qos trust cos
      switch(config-if-Et17)#interface ethernet 18
      switch(config-if-Et18)#switchport
      switch(config-if-Et18)#default qos trust
      switch(config-if-Et19)#interface ethernet 19
      switch(config-if-Et19)#no switchport
      switch(config-if-Et19)#default qos trust
      switch(config-if-Et19)#show qos interface ethernet 15 - 19 trust
      Port Trust Mode
       Operational Configured
      ---------------------------------------------------------------
      Ethernet15 DSCPDSCP
      Ethernet16 UNTRUSTED UNTRUSTED
      Ethernet17 COS COS
      Ethernet18 COS DEFAULT
      Ethernet19 DSCPDEFAULT
      
      switch(config-if-Et19)#

Configuring Default Port Settings

Default CoS and DSCP settings are assigned to individual port channel and Ethernet interfaces. These configuration mode interface commands specify the port’s default CoS and DSCP values.

  • qos cos configures a port’s default CoS value.
  • qos dscp configures a port’s default DSCP value.

Example

These commands configure default CoS (4) and DSCP (44) settings on Ethernet interface 19.
switch(config)#interface ethernet 19
switch(config-if-Et19)#qos cos 4
switch(config-if-Et19)#qos dscp 44
switch(config-if-Et19)#show active
interface Ethernet19
 qos cos 4
 qos dscp 44
switch(config-if-Et19)#show qos interfaces ethernet 19
Ethernet19:
 Trust Mode: COS
 Default COS: 4
 Default DSCP: 44

switch(config-if-Et19)#

Traffic Class Derivations – FM6000 Platform Switches

Traffic Classesdescribes traffic classes.

Traffic Class Derivation Source

The following table displays the source for deriving a data stream’s traffic class.

Table 11. Traffic Class Derivation Source: FM6000 Platform Switches
  Untrusted CoS Trusted DSCP Trusted
Untagged Non-IP Default CoS (port) Default CoS (port) Default DSCP (port)
Untagged IP Default CoS (port) Default CoS (port) DSCP (packet)
Tagged Non-IP Default CoS (port) CoS (packet) Default DSCP (port)
Tagged IP Default CoS (port) CoS (packet) DSCP (packet)

CoS and DSCP Port Settings – FM6000 Platform Switchesdescribes the default CoS and DSCP settings for each port.

Mapping CoS to Traffic Class

The qos map cos command assigns a traffic class to a list of CoS settings. Multiple commands create a complete CoS to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s CoS field or the port upon which it is received.

Example

This command assigns the traffic class of 5 to the classes of service 1, 3, 5, and 7.
switch(config)#qos map cos 1 3 5 7 to traffic-class 5
switch(config)#show qos maps
   Number of Traffic Classes supported: 8
   Number of Transmit Queues supported: 8


   Cos-tc map:
     cos:01234567
     ----------------------------
     tc: 15254565


switch(config)#

The following table displays the default CoS to Traffic Class map on FM6000 platform switches.

Table 12. Default CoS to Traffic Class Map: FM6000 Platform Switches
Inbound CoS Untagged 0 1 2 3 4 5 6 7
Traffic Class Derived: use default CoS as inbound CoS 1 0 2 3 4 5 6 7

Mapping DSCP to Traffic Class

The qos map dscp command assigns a traffic class to a set of DSCP values. Multiple commands create a complete DSCP to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s DSCP field or the chip upon which it is received.

Example

This command assigns the traffic class of three to the DSCP values of 12, 13, 25, and 37.
switch(config)#qos map dscp 12 13 25 37 to traffic-class 3
switch(config)#show qos map
 Number of Traffic Classes supported: 8


 Dscp-tc map:
 d1 :d2 0123456789
 --------------------------------------
0 : 1111111100
1 : 0033002222
2 : 2222333333
3 : 3344444344
4 : 5555555566
5 : 6666667777
6 : 7777


switch(config)#

The following displays the default DSCP to Traffic Class map on FM6000 platform switches.

Table 13. Default DSCP to Traffic Class Map: FM6000 Platform Switches
Inbound DSCP 0-7 8-15 16-23 24-31 32-39 40-47 48-55 56-63
Traffic Class 1 0 2 3 4 5 6 7

CoS and DSCP Rewrite – FM6000 Platform Switches

Rewriting CoS and DSCP describes the CoS and DSCP rewrite functions.

Traffic Class to CoS Rewrite Map

The CoS rewrite value is configurable and based on a data stream’s traffic class, as specified by the traffic class-CoS rewrite map. The qos map traffic-class to cos command assigns a CoS rewrite value to a list of traffic classes. Multiple commands create the complete traffic class–CoS rewrite map.

Example

This command assigns the CoS rewrite value of two to traffic classes 1, 3, and 5.
switch(config)#qos map traffic-class 1 3 5 to cos 2
switch(config)#show qos map
 Number of Traffic Classes supported: 8


 Tc - tx-queue map:
 tc:01234567
 ---------------------------------
 tx-queue:01234567


switch(config)#

The following table displays the default traffic class–CoS rewrite map on FM6000 platform switches.

Table 14. Default Traffic Class to CoS Rewrite Map: FM6000 Platform Switches

Traffic Class

0

1

2

3

4

5

6

7

CoS Rewrite Value

1

0

2

3

4

5

6

7

Traffic Class to DSCP Rewrite Map

The DSCP rewrite value is configurable and based on a data stream’s traffic class, as specified by the traffic class-DSCP rewrite map. The qos map traffic-class to dscp command assigns a DSCP rewrite value to a list of traffic classes. Multiple commands create the complete traffic class-DSCP rewrite map.

Example

This command assigns the DSCP rewrite value of 37 to traffic classes 2, 4, and 6.
switch(config)#qos map traffic-class 2 4 6 to dscp 37
switch(config)#show qos map
 Number of Traffic Classes supported: 8


 Tc-dscp map:
 tc:01234567
 -----------------------------
 dscp:80 37 24 37 40 37 56


switch(config)#

The following table displays the default traffic class–DSCP rewrite map on on FM6000 platform switches.

Table 15. Default Traffic Class to DSCP Rewrite Map: FM6000 Platform Switches

Traffic Class

0

1

2

3

4

5

6

7

DSCP Rewrite Value

8

0

16

24

32

40

48

56

Transmit Queues and Port Shaping – FM6000 Platform Switches

Transmit Queues and Port Shaping describes transmit queues and port shaping.

A data stream’s traffic class determines the transmit queue it uses. The switch defines a single traffic class-transmit queue map for all Ethernet and port channel interfaces and is used for unicast and multicast traffic. The show qos maps command displays the traffic class to transmit queue map.

The following table displays the default traffic class to transmit queue map on FM6000 platform switches.

Table 16. Default Traffic Class to Transmit Queue Map: FM6000 Platform Switches
Traffic Class 0 1 2 3 4 5 6 7
Transmit Queue 0 1 2 3 4 5 6 7

Mapping Traffic Classes to a Transmit Queue

The qos map traffic-class to tx-queue command assigns traffic classes to a transmit queue. Multiple commands create the complete map.

Example

These commands assign traffic classes of 1, 3, and 5 to transmit queue 1, traffic classes 2, 4, and 6 to transmit queue 2, and traffic class 0 to transmit queue 0, then display the resultant map.
switch(config)#qos map traffic-class 1 3 5 to tx-queue 1
switch(config)#qos map traffic-class 2 4 6 to tx-queue 2
switch(config)#qos map traffic-class 0 to tx-queue 0
switch(config)#show qos maps
   Number of Traffic Classes supported: 8
   Number of Transmit Queues supported: 8


   Tc - tx-queue map:
     tc:        01234567
 ---------------------------------
     tx-queue:  01212127

switch(config)#

Entering TX-Queue Configuration Mode

Transmit queues are configurable on Ethernet ports and port channels. Queue parameters are configured in tx-queue configuration command mode, which is entered from interface ethernet configuration mode. The tx-queue (FM6000) command places the switch in tx-queue configuration mode. The show qos interfaces displays the transmit queue configuration for a specified port.

Example

This command enters tx-queue configuration mode for transmit queue 3 of Ethernet interface 5.
switch(config)#interface ethernet 5
switch(config-if-Et5)#tx-queue 3 
switch(config-if-Et5-txq-3)#

Configuring the Shape Rate – Port and Transmit Queues

A port’s shape rate specifies its maximum outbound traffic bandwidth. A transmit queue’s shape rate specifies the queue’s maximum outbound bandwidth. Shape rate commands specify data rates in kbps.

Note: Enabling port shaping on an FM6000 interface disables queue shaping internally. Disabling port shaping restores queue shaping as specified in running-config.

Example

These commands configure a shape rate of 5 Gbs on Ethernet port 3, then configure the shape rate for the following transmit queues:
    • transmit queues 0, 1, and 2: 500 Mbps
    • transmit queues 3, 4, and 5: 400 Mbps

      switch(config)#interface ethernet 3
      switch(config-if-Et3)#shape rate 5000000
      switch(config-if-Et3)#tx-queue 0
      switch(config-if-Et3-txq-0)#shape rate 500000
      switch(config-if-Et3-txq-0)#tx-queue 1
      switch(config-if-Et3-txq-1)#shape rate 500000
      switch(config-if-Et3-txq-1)#tx-queue 3
      switch(config-if-Et3-txq-3)#shape rate 400000
      switch(config-if-Et3-txq-3)#tx-queue 4
      switch(config-if-Et3-txq-4)#shape rate 400000
      switch(config-if-Et3-txq-4)#tx-queue 5
      switch(config-if-Et3-txq-5)#shape rate 400000
      switch(config-if-Et3-txq-5)#exit
      switch(config-if-Et3)#show qos interface ethernet 3
      Ethernet3:
      
      
       Port shaping rate: 5000000Kbps
      
       Tx-Queue BandwidthShape Rate Priority
      (percent) (Kbps)
       -----------------------------------------------
      7 N/A disabledstrict
      6 N/A disabledstrict
      5 N/A 400000strict
      4 N/A 400000strict
      3 N/A 400000strict
      2 N/A disabledstrict
      1 N/A 500000strict
      0 N/A 500000strict
      
      switch(config-if-Et3)#

Configuring Queue Priority

Queue priority rank is denoted by the queue number; transmit queues with higher numbers have higher priority. The priority (FM6000) command configures a transmit queue’s priority type:

  • priority strict configures the queue as a strict priority queue.
  • no priority configures the queue as a round robin queue.

A queue’s configuration as round robin also applies to all lower priority queues regardless of other configuration statements.

The bandwidth percent (FM6000) command configures a round robin queue’s bandwidth share. The cumulative operational bandwidth of all round robin queues is always less than or equal to 100%. If the cumulative configured bandwidth is greater than 100%, each port’s operational bandwidth is its configured bandwidth divided by the cumulative configured bandwidth.

Examples
  • These commands configure transmit queue 3 (on Ethernet interface 19) as a round robin queue, then allocates 10%, 20%, 30%, and 40% bandwidth to queues 0 through 3.
    The no priority statement for queue 3 also configures queues 0, 1, and 2 as round robin queues. Removing this statement reverts the other queues to strict priority type unless running-config contains a no priority statement for one of these queues.
    switch(config)#interface ethernet 19
    switch(config-if-Et19)#tx-queue 3
    switch(config-if-Et19-txq-3)#no priority
    switch(config-if-Et19-txq-3)#bandwidth percent 40
    switch(config-if-Et19-txq-3)#tx-queue 2
    switch(config-if-Et19-txq-2)#bandwidth percent 30
    switch(config-if-Et19-txq-2)#tx-queue 1
    switch(config-if-Et19-txq-1)#bandwidth percent 20
    switch(config-if-Et19-txq-1)#tx-queue 0
    switch(config-if-Et19-txq-0)#bandwidth percent 10
    switch(config-if-Et19-txq-0)#show qos interface ethernet 19
    Ethernet19:
    
     Port shaping rate: disabled
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
     -----------------------------------------------
    7 N/A disabledstrict
    6 N/A disabledstrict
    5 N/A disabledstrict
    4 N/A disabledstrict
    340 disabled round-robin
    230 disabled round-robin
    120 disabled round-robin
    010 disabled round-robin
    
    switch(config-if-Et19-txq-0)#
  • Changing the bandwidth percentage for queue 3 to 60 changes the operational bandwidth of each queue to its configured bandwidth divided by 120% (10%+20%+30%+60%).
    switch(config-if-Et19-txq-0)#tx-queue 3
    switch(config-if-Et19-txq-3)#bandwidth percent 60
    switch(config-if-Et19-txq-3)#show qos interface ethernet 19
    Ethernet19:
    
     Port shaping rate: disabled
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
     -----------------------------------------------
    7 N/A disabledstrict
    6 N/A disabledstrict
    5 N/A disabledstrict
    4 N/A disabledstrict
    349 disabled round-robin
    224 disabled round-robin
    116 disabled round-robin
    0 8 disabled round-robin
    
    switch(config-if-Et19-txq-3)#

QoS Configuration: Petra Platform Switches

Implementing QoS on a Petra platform switch consists of configuring port trust settings, default port settings, default traffic classes, conversion maps, and transmit queues.

CoS and DSCP Port Settings – Petra Platform Switches

Port Settings – Trust Mode and Traffic Classdescribes port trust and default port CoS and DSCP values.

Configuring Port Trust Settings

The qos trust command configures the QoS port trust mode for the configuration mode interface. Trust enabled ports use packet CoS or DSCP values to classify traffic. The port-trust default for switched ports is cos. The port-trust default for routed ports is dscp.

  • qos trust cos specifies cos as the port’s port-trust mode.
  • qos trust dscp specifies dscp as the port’s port-trust mode.
  • no qos trust specifies untrusted as the port’s port-trust mode.

The show qos interfaces trust command displays the trust mode of specified interfaces.

Example

  • These commands configure and display the following trust modes:

    • Ethernet 3/25: dscp
    • Ethernet 3/26: untrusted
    • Ethernet 3/27: cos
    • Ethernet 3/28: default as a switched port
    • Ethernet 3/29: default as a routed port

      switch(config)#interface ethernet 3/25
      switch(config-if-Et3/25)#qos trust dscp
      switch(config-if-Et3/25)#interface ethernet 3/26
      switch(config-if-Et3/26)#no qos trust
      switch(config-if-Et3/26)#interface ethernet 3/27
      switch(config-if-Et3/27)#qos trust cos
      switch(config-if-Et3/27)#interface ethernet 3/28
      switch(config-if-Et3/28)#switchport
      switch(config-if-Et3/28)#default qos trust
      switch(config-if-Et3/28)#interface ethernet 3/29
      switch(config-if-Et3/29)#no switchport
      switch(config-if-Et3/29)#default qos trust
      switch(config-if-Et3/29)#show qos interface ethernet 3/25 - 3/29 trust
      Port Trust Mode
       Operational Configured
      ---------------------------------------------------------------
      Ethernet3/25 DSCPDSCP
      Ethernet3/26 UNTRUSTED UNTRUSTED
      Ethernet3/27 COS COS
      Ethernet3/28 COS DEFAULT
      Ethernet3/29 DSCPDEFAULT
      
      switch(config-if-Et3/29)#

Configuring Default Port Settings

Port channel and Ethernet interfaces are not assigned default CoS or DSCP settings.

Traffic Class Derivations – Petra Platform Switches

Traffic Classes describes traffic classes.

Traffic Class Derivation Source

The following table displays the source for deriving a data stream’s default traffic class.

Table 17. Traffic Class Derivation Source: Petra Platform Switches
  Untrusted CoS Trusted DSCP Trusted
Untagged Non-IP Default TC (chip) Default TC (chip) Default TC (chip)
Untagged IP Default TC (chip) Default TC (chip) DSCP (packet)
Tagged Non-I Default TC (chip) CoS (packet) Default TC (chip)
Tagged IP Default TC (chip) CoS (packet) DSCP (packet)

Configuring Default Traffic Class

Petra platform switches assign a default traffic class to the set of Ethernet interfaces controlled by individual PetraA chips. Default traffic class values are configurable for each PetraA chip, not individual interfaces.

The platform petraA traffic-class command specifies the default traffic class used by all ports controlled by a specified chip. The show platform petraA traffic-classcommand displays traffic class assignments.

Examples
  • This command configures the default traffic class to five for the ports 32-39 on linecard 3 (7500 Series).
    switch(config)#platform petraA petra3/4 traffic-class 5
    switch(config)#show platform petraA module 3 traffic-class
    Petra3/0 traffic-class: 1
    Petra3/1 traffic-class: 1
    Petra3/2 traffic-class: 1
    Petra3/3 traffic-class: 1
    Petra3/4 traffic-class: 5
    Petra3/5 traffic-class: 1
    switch(config)#
  • This command configures the default traffic class to three for all ports on linecard 6 (7500 Series).
    switch(config)#platform petraA module 6 traffic-class 6
    switch(config)#show platform petraA module 6 traffic-class
    Petra6/0 traffic-class: 6
    Petra6/1 traffic-class: 6
    Petra6/2 traffic-class: 6
    Petra6/3 traffic-class: 6
    Petra6/4 traffic-class: 6
    Petra6/5 traffic-class: 6
    switch(config)#

Mapping CoS to Traffic Class

The qos map cos command assigns a traffic class to a list of CoS settings. Multiple commands create a complete CoS–traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s CoS field or the port upon which it is received.

Example

This command assigns the traffic class of 4 to the classes of service 1, 3, 5, and 7.
switch(config)#qos map cos 1 3 5 7 to traffic-class 4
switch(config)#show qos maps
 Number of Traffic Classes supported: 8


 Cos-tc map:
 cos:01234567
 ----------------------------
 tc: 14244464


switch(config)#

The following table displays the default CoS to traffic class map on Petra platform switches.

Table 18. Default CoS to Traffic Class Map: Petra Platform Switches
Inbound CoS untagged 0 1 2 3 4 5 6 7
Traffic Class Derived: use default CoS as inbound CoS 1 0 2 3 4 5 6 7

Mapping DSCP to Traffic Class

The qos map dscp command assigns a traffic class to a set of DSCP values. Multiple commands create a complete DSCP to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s DSCP field or the chip upon which it is received.

Example

This command assigns the traffic class of three to the DSCP values of 12, 13, 25, and 37.
switch(config)#qos map dscp 12 13 14 25 48 to traffic-class 3
switch(config)#show qos maps
 Number of Traffic Classes supported: 8


 Dscp-tc map:
 d1 :d2 0123456789
 --------------------------------------
0 : 1111111100
1 : 0033302222
2 : 2222333333
3 : 3344444444
4 : 5555555536
5 : 6666667777
6 : 7777


switch(config)#

The following table displays the default DSCP to Traffic Class map on Petra platform switches.

Table 19. Default DSCP to Traffic Class Map: Petra Platform Switches
Inbound DSCP 0-7 8-15 16-23 24-31 32-39 40-47 48-55 56-63
Traffic Class 1 0 2 3 4 5 6 7

CoS Rewrite – Petra Platform Switches

Rewriting CoS and DSCP describes the CoS rewrite function.

Traffic Class to CoS Rewrite Map

The CoS rewrite value is configurable and based on a data stream’s traffic class, as specified by the traffic class-CoS rewrite map. The qos map traffic-class to coscommand assigns a CoS rewrite value to a list of traffic classes. Multiple commands create the complete traffic class-CoS rewrite map.

Example

This command assigns the CoS of two to traffic classes 1, 3, and 5.
switch(config)#qos map traffic-class 1 3 5 to cos 2
switch(config)#show qos map
 Number of Traffic Classes supported: 8

 Tc-cos map:
 tc: 01234567
 ----------------------------
 cos:12224267

switch(config)#

The following table displays the default Traffic Class to CoS rewrite value map on Petra platform switches.

Table 20. Default Traffic Class to CoS Rewrite Value Map: Petra Platform Switches

Traffic Class

0

1

2

3

4

5

6

7

CoS Rewrite Value

1

0

2

3

4

5

6

7

Traffic Class to DSCP Rewrite Map

DSCP rewrite is always disabled on Petra platform switches.

Transmit Queues and Port Shaping – Petra Platform Switches

Transmit Queues and Port Shaping describes transmit queues and port shaping.

Petra platform switches provide four physical queues for each egress port: Unicast High, Unicast Low, Multicast High, and Multicast Low. Data is scheduled for the high or low queue based on its priority as defined by its transmit queue assignment (unicast traffic) or traffic class (multicast traffic), as shown in the table below. A Petra transmit queue is a data structure that defines scheduling of unicast traffic among phyical egress queues.

Table 21. Traffic Distribution to Egress Port Queues
 

High Priority Queue

Low Priority Queue

Unicast Traffic

Transmit Queues 5 – 7

Transmit Queues 0 – 4

Multicast Traffic

Traffic Classes 5 – 7

Traffic Classes 0 – 4

Multicast queue capacity that is available after multicast traffic is serviced is used for unicast traffic of a corresponding priority. Similarly, unicast queue capacity that is available after unicast traffic is serviced is used for overflow multicast traffic. Under conditions of unicast and multicast congestion, egress traffic is evenly split between unicast and multicast traffic.

Unicast Transmit Queues and Port Shapingdescribes unicast transmit queues and shaping. Multicast Egress Scheduling describes multicast priority and traffic classes.

Unicast Transmit Queues and Port Shaping

A data stream’s traffic class determines the transmit queue it uses. The switch defines a single traffic class–transmit queue map for unicast traffic on all Ethernet interfaces. The show qos maps command displays the traffic class–transmit queue map. The following table displays the default traffic class to transmit queue map on Petra platform switches.

Table 22. Default Traffic Class to Transmit Queue Map: Petra Platform Switches
Traffic Class 0 1 2 3 4 5 6 7
Transmit Queue 0 1 2 3 4 5 6 7

Transmit queue parameters are configured in tx-queue configuration command mode.

Mapping Traffic Classes to a Transmit Queue

The qos map traffic-class to tx-queue command assigns traffic classes to a transmit queue. Multiple commands complete the traffic class-transmit queue map. Traffic class 7 and transmit queue 7 are always mapped to each other. This association is not editable.

Example

These commands assign traffic classes of 1, 3, and 5 to transmit queue 1, traffic classes 2, 4, and 6 to transmit queue 2, and traffic class 0 to transmit queue 0, then display the resultant map.
switch(config)#qos map traffic-class 1 3 5 to tx-queue 1
switch(config)#qos map traffic-class 2 4 6 to tx-queue 2
switch(config)#qos map traffic-class 0 to tx-queue 0
switch(config)#show qos maps
 Number of Traffic Classes supported: 8
 Number of Transmit Queues supported: 8

 Tc - tx-queue map:
 tc:01234567
 ---------------------------------
 tx-queue:01212127

switch(config)#
Entering Tx-Queue Configuration Mode

The tx-queue (Petra) command places the switch in tx-queue configuration mode to configure a transmit queue on the configuration mode interface. Tx-queue 7 is not configurable. The show qos interfaces displays the transmit queue configuration for a specified port.

Example

This command enters tx-queue configuration mode for transmit queue 3 of Ethernet interface 3/28.
switch(config)#interface ethernet 3/28
switch(config-if-Et3/28)#tx-queue 3
switch(config-if-Et3/28-txq-3)#
Configuring the Shape Rate – Port and Transmit Queues

A port’s shape rate specifies its maximum outbound traffic bandwidth. A transmit queue’s shape rate specifies the queue’s maximum outbound bandwidth. Shape rate commands specify data rates in kbps.

Example

These commands configure a shape rate of 5 Gbs on Ethernet port 3, then configure the shape rate for the following transmit queues:
  • transmit queues 0, 1, and 2: 500 Mbps
  • transmit queues 3, 4, and 5: 400 Mbps

    switch(config)#interface ethernet 3/28
    switch(config-if-Et3/28)#shape rate 5000000
    switch(config-if-Et3/28)#tx-queue 0
    switch(config-if-Et3/28-txq-0)#shape rate 500000
    switch(config-if-Et3/28-txq-0)#tx-queue 1
    switch(config-if-Et3/28-txq-1)#shape rate 500000
    switch(config-if-Et3/28-txq-1)#tx-queue 2
    switch(config-if-Et3/28-txq-2)#shape rate 500000
    switch(config-if-Et3/28-txq-5)#tx-queue 3
    switch(config-if-Et3/28-txq-3)#shape rate 400000
    switch(config-if-Et3/28-txq-3)#tx-queue 4
    switch(config-if-Et3/28-txq-4)#shape rate 400000
    switch(config-if-Et3/28-txq-4)#tx-queue 5
    switch(config-if-Et3/28-txq-5)#shape rate 400000
    switch(config-if-Et3/28-txq-5)#show qos interface ethernet 3/28
    Ethernet3/28:
    
     Port shaping rate: 5000000Kbps
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
     -----------------------------------------------
    7 N/A disabledstrict
    6 N/A disabledstrict
    5 N/A 400000strict
    4 N/A 400000strict
    3 N/A 400000strict
    2 N/A 500000strict
    1 N/A 500000strict
    0 N/A 500000strict
    
    switch(config-if-Et3/28-txq-5)#
Configuring Queue Priority

The priority (Petra) command configures a transmit queue’s priority type:

  • The priority strict command configures the queue as a strict priority queue.
  • The no priority command configures the queue as a round robin queue.

A queue’s configuration as round robin also applies to all lower priority queues regardless of other configuration statements.

The bandwidth percent (Petra) command configures a round robin queue’s bandwidth share. The cumulative operational bandwidth of all round robin queues is always less than or equal to 100%. If the cumulative configured bandwidth is greater than 100%, each port’s operational bandwidth is its configured bandwidth divided by the cumulative configured bandwidth.

Examples
  • These commands configure transmit queue 3 (on Ethernet interface 3/28) as a round robin queue, then allocates 10%, 20%, 30%, and 40% bandwidth to queues 0 through 3.
    The no priority statement for queue 3 also configures queues 0, 1, and 2 as round robin queues. Removing this statement reverts the other queues to strict priority type unless running-config contains a no priority statement for one of these queues.
    switch(config-if-Et3/28)#tx-queue 3
    switch(config-if-Et3/28-txq-3)#no priority
    switch(config-if-Et3/28-txq-3)#bandwidth percent 40
    switch(config-if-Et3/28-txq-3)#tx-queue 2
    switch(config-if-Et3/28-txq-2)#bandwidth percent 30
    switch(config-if-Et3/28-txq-2)#tx-queue 1
    switch(config-if-Et3/28-txq-1)#bandwidth percent 20
    switch(config-if-Et3/28-txq-1)#tx-queue 0
    switch(config-if-Et3/28-txq-0)#bandwidth percent 10
    switch(config-if-Et3/28-txq-0)#show qos interface ethernet 3/28
    Ethernet3/28:
    
     Port shaping rate: 5000000Kbps
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
     -----------------------------------------------
    7 N/A disabledstrict
    6 N/A disabledstrict
    5 N/A 400000strict
    4 N/A 400000strict
    340 400000 round-robin
    230 500000 round-robin
    120 500000 round-robin
    010 500000 round-robin
    
    switch(config-if-Et3/28-txq-0)#
  • Changing the bandwidth percentage for queue 3 to 60 changes the operational bandwidth of each queue to its configured bandwidth divided by 120% (10%+20%+30%+60%).
    switch(config-if-Et3/28-txq-0)#tx-queue 3
    switch(config-if-Et3/28-txq-3)#bandwidth percent 60
    switch(config-if-Et3/28-txq-3)#show qos interface ethernet 3/28
    Ethernet3/28:
    
     Port shaping rate: 5000000Kbps
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
     -----------------------------------------------
    7 N/A disabledstrict
    6 N/A disabledstrict
    5 N/A 400000strict
    4 N/A 400000strict
    349 400000 round-robin
    224 500000 round-robin
    116 500000 round-robin
    0 8 500000 round-robin
    switch(config-if-Et3/28-txq-3)#

Multicast Egress Scheduling

Multiclass traffic is not affected by traffic class assignment or port shaping statements. Multicast traffic is assigned to port egress queues based on traffic class and uses strict priority to schedule egress between the high and low queues.

 

QoS Configuration: Trident and Tomahawk Platform Switches

CoS and DSCP Port Settings – Trident and Tomahawk Platform Switches

Configuring Port Trust Settings

The qos trust command configures the QoS port trust mode for the configuration mode interface. Trust-enabled ports use packet CoS or DSCP values to classify traffic. The port-trust default for switched ports is CoS. The port-trust default for routed ports is DSCP.

  • qos trust cos specifies CoS as the port’s trust mode.
  • qos trust dscp specifies DSCP as the port’s trust mode.
  • no qos trust specifies untrusted as the port’s trust mode.

The show qos interfaces trust command displays the trust mode of specified interfaces.

Example

  • These commands configure and display the following trust modes:

    • Ethernet 15: dscp
    • Ethernet 16: untrusted
    • Ethernet 17: cos
    • Ethernet 18: default as a switched port
    • Ethernet 19: default as a routed port
      switch(config)#interface ethernet 15
      switch(config-if-Et15)#qos trust dscp
      switch(config-if-Et15)#interface ethernet 16
      switch(config-if-Et16)#no qos trust
      switch(config-if-Et16)#interface ethernet 17
      switch(config-if-Et17)#qos trust cos
      switch(config-if-Et17)#interface ethernet 18
      switch(config-if-Et18)#switchport
      switch(config-if-Et18)#default qos trust
      switch(config-if-Et18)#interface ethernet 19
      switch(config-if-Et19)#no switchport
      switch(config-if-Et19)#default qos trust
      switch(config-if-Et19)#show qos interface ethernet 15 - 19 trust
      Port Trust Mode
       Operational Configured
      ---------------------------------------------------------------
      Ethernet15 DSCPDSCP
      Ethernet16 UNTRUSTED UNTRUSTED
      Ethernet17 COS COS
      Ethernet18 COS DEFAULT
      Ethernet19 DSCPDEFAULT
      
      switch(config-if-Et19)#

Configuring Default Port Settings

Default CoS and DSCP settings are assigned to individual port channel and Ethernet interfaces. These configuration mode interface commands specify the port’s default CoS and DSCP values.

  • qos cos configures a port’s default CoS value.
  • qos dscp configures a port’s default DSCP value.

Example

These commands configure default CoS (4) and DSCP (44) values on Ethernet interface 7.
switch(config)#interface ethernet 7
switch(config-if-Et7)#qos cos 4
switch(config-if-Et7)#qos dscp 44
switch(config-if-Et7)#show active
interface Ethernet7
 qos cos 4
 qos dscp 44
switch(config-if-Et7)#show qos interfaces ethernet 7
Ethernet7:
 Trust Mode: COS
 Default COS: 4
 Default DSCP: 44

switch(config-if-Et7)#

Traffic Class Derivations – Trident and Tomahawk Platform Switches

Traffic Classesdescribes traffic classes.

Traffic Class Derivation Source

The following table displays the source for deriving a data stream’s traffic class.

Table 23. Traffic Class Derivation Source: Trident and Tomahawk Platform Switches
  Untrusted CoS Trusted DSCP Trusted
Untagged Non-IP Default CoS (port) Default CoS (port) Default DSCP (port)
Untagged IP Default CoS (port) Default CoS (port) DSCP (packet)
Tagged Non-IP Default CoS (port) CoS (packet) Default DSCP (port)
Tagged IP Default CoS (port) CoS (packet) DSCP (packet)

CoS and DSCP Port Settings – Trident and Tomahawk Platform Switches describes the default CoS and DSCP settings for each port.

Mapping CoS to Traffic Class

The qos map cos command assigns a traffic class to a list of CoS settings. Multiple commands create a complete CoS to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s CoS field or the port upon which it is received.

Example

This command assigns the traffic class of 5 to the classes of service 1, 3, 5, and 7.
switch(config)#qos map cos 1 3 5 7 to traffic-class 5
switch(config)#show qos maps
 Number of Traffic Classes supported: 8

 Cos-tc map:
 cos:01234567
 ----------------------------
 tc: 15254565

switch(config)#

The following table displays the default CoS–traffic class map on Trident and Tomahawk platform switches.

Table 24. Default CoS to Traffic Class Map: Trident and Tomahawk Platform Switches
Inbound CoS 0 1 2 3 4 5 6 7
Traffic Class 1 0 2 3 4 5 6 7

Mapping DSCP to Traffic Class

The qos map dscp command assigns a traffic class to a set of DSCP values. Multiple commands create a complete DSCP to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s DSCP field or the chip upon which it is received.

Example

This command assigns the traffic class of 0 to DSCP values of 12, 24, 41, and 44-47.
switch(config)#qos map dscp 12 24 41 44 45 46 47 to traffic-class 0
switch(config)#show qos maps
 Number of Traffic Classes supported: 8

 Dscp-tc map:
 d1 :d2 0123456789
 --------------------------------------
0 : 1111111100
1 : 0000002222
2 : 2222033333
3 : 3344444444
4 : 5055000066
5 : 6666667777
6 : 7777

switch(config)#

The following table displays the default DSCP–traffic class map on Trident and Tomahawk platform switches.

Table 25. Default DSCP to Traffic Class Map: Trident and Tomahawk Platform Switches
Inbound DSCP 0-7 8-15 16-23 24-31 32-39 40-47 48-55 56-63
Traffic Class 1 0 2 3 4 5 6 7

CoS and DSCP Rewrite – Trident and Tomahawk Platform Switches

Rewriting CoS and DSCPdescribes the CoS and DSCP rewrite functions.

Traffic Class to CoS Rewrite Map

The CoS rewrite value is configurable and based on a data stream’s traffic class, as specified by the traffic class-CoS rewrite map. The qos map traffic-class to cos command assigns a CoS rewrite value to a list of traffic classes. Multiple commands create the complete traffic class–CoS rewrite map.

Example

This command assigns the CoS of two to traffic classes 1, 3, and 5.
switch(config)#qos map traffic-class 1 3 5 to cos 2
switch(config)#show qos map
 Number of Traffic Classes supported: 8

 Tc-cos map:
 tc: 01234567
 ----------------------------
 cos:12224267

switch(config)#

The following table displays the default Traffic Class to CoS rewrite value map on Trident and Tomahawk platform switches.

Table 26. Default Traffic Class to CoS Rewrite Value Map: Trident and Tomahawk Platform Switches

Traffic Class

0

1

2

3

4

5

6

7

CoS Rewrite Value

1

0

2

3

4

5

6

7

Traffic Class to DSCP Rewrite Map

The DSCP rewrite value is configurable and based on a data stream’s traffic class, as specified by the traffic class-DSCP rewrite map. The qos map traffic-class to dscp command assigns a DSCP rewrite value to a list of traffic classes. Multiple commands create the complete traffic class-DSCP rewrite map.

Example

This command assigns the DSCP value of 29 to traffic classes 2, 4, and 6.
switch(config)#qos map traffic-class 2 4 6 to dscp 29
switch(config)#show qos map
 Number of Traffic Classes supported: 8

 Tc-dscp map:
 tc:01234567
 -----------------------------
 dscp:80 29 24 29 40 29 56

switch(config)#

The following displays the default traffic class to DSCP rewrite map on Trident and Tomahawk platform switches.

Table 27. Traffic Class to DSCP Rewrite Value Map: Trident and Tomahawk Platform Switches

Traffic Class

0

1

2

3

4

5

6

7

DSCP

8

0

16

24

32

40

48

56

Transmit Queues and Port Shaping – Trident and Tomahawk Platform Switches

Transmit Queues and Port Shaping describes transmit queues and port shaping.

Trident and Tomahawk platform switches define 12 transmit queues: eight unicast (UC0 – UC7) and four multicast (MC0 – MC03). The traffic class–transmit queue maps are configured globally and apply to all Ethernet interfaces. The show qos maps command displays the traffic class–transmit queue maps.

The following table displays the default traffic class–transmit queue maps.

Table 28. Default Traffic Class to Transmit Queue Map: Trident and Tomahawk Platform Switches
Traffic Class 0 1 2 3 4 5 6 7
Unicast Transmit Queue 0 1 2 3 4 5 6 7
Multicast Transmit Queue 0 0 1 1 2 2 3 3

Mapping Traffic Classes to a Transmit Queue

These commands assign traffic classes to a transmit queue:

Multiple commands create the complete maps.

Example
  • These commands assign the following on Ethernet interface 7:

    • traffic classes 1, 3, and 5 to unicast queue 1
    • traffic classes 2, 4, and 6 to unicast queue 5
    • traffic classes 1, 2, and 3 to multicast queue 1
    • traffic classes 4, 5, and 6 to multicast queue 3
    • traffic class 0 to unicast queue 0 and multicast queue 0
      switch(config)#default interface ethernet 7
      switch(config)#qos map traffic-class 1 3 5 to uc-tx-queue 1
      switch(config)#qos map traffic-class 2 4 6 to uc-tx-queue 5
      switch(config)#qos map traffic-class 1 2 3 to mc-tx-queue 1
      switch(config)#qos map traffic-class 4 5 6 to mc-tx-queue 3
      switch(config)#qos map traffic-class 0 to uc-tx-queue 0
      switch(config)#qos map traffic-class 0 to mc-tx-queue 0
      switch(config)#show qos maps
       Number of Traffic Classes supported: 8
       Number of Transmit Queues supported: 12
      
       Tc - uc-tx-queue map:
       tc: 01234567
       ------------------------------------
       uc-tx-queue:01515157
      
       Tc - mc-tx-queue map:
       tc: 01234567
       ------------------------------------
       mc-tx-queue:01113333
      
      switch(config)#

Entering a Transmit Queue Configuration Mode

Transmit queues are configurable on individual Ethernet ports. Parameters for individual transmit queues are configured in one of two transmit queue configuration modes. Transmit queue modes are accessed from an interface-ethernet configuration mode.

  • uc-tx-queue places the switch in uc-tx-queue mode to configure a unicast transmit queue.
  • mc-tx-queue places the switch in mc-tx-queue mode to configure a multicast transmit queue.

The show qos interfaces displays the transmit queue configuration for a specified port.

Examples
  • This command enters the mode that configures unicast transmit queue 3 of Ethernet interface 5.
    switch(config)#interface ethernet 5
    switch(config-if-Et5)#uc-tx-queue 3
    switch(config-if-Et5-uc-txq-3)#
  • This command enters the mode to configure multicast transmit queue 3 of Ethernet interface 5.
    switch(config-if-Et5)#mc-tx-queue 2
    switch(config-if-Et5-mc-txq-2)#

Configuring the Shape Rate – Port and Transmit Queues

A port’s shape rate specifies the port’s maximum outbound traffic bandwidth. A shape rate can also be configured for all transmit queues on each port. All shape rate commands use kbps to specify data rates.

Example

These commands configure a shape rate of 5 Gbs on Ethernet port 7, then configure the shape rate for the following transmit queues:
    • unicast transmit queues 0 and 1: 500 Mbps
    • unicast transmit queues 3 and 4: 400 Mbps
    • multicast transmit queues 0 and 2: 300 Mbps

      switch(config)#interface ethernet 7
      switch(config-if-Et7)#shape rate 5000000
      switch(config-if-Et7)#uc-tx-queue 0
      switch(config-if-Et7-uc-txq-0)#shape rate 500000
      switch(config-if-Et7-uc-txq-0)#uc-tx-queue 1
      switch(config-if-Et7-uc-txq-1)#shape rate 500000
      switch(config-if-Et7-uc-txq-1)#uc-tx-queue 3
      switch(config-if-Et7-uc-txq-3)#shape rate 400000
      switch(config-if-Et7-uc-txq-3)#uc-tx-queue 5
      switch(config-if-Et7-uc-txq-5)#shape rate 400000
      switch(config-if-Et7-uc-txq-5)#mc-tx-queue 0
      switch(config-if-Et7-mc-txq-0)#shape rate 300000
      switch(config-if-Et7-mc-txq-0)#mc-tx-queue 2
      switch(config-if-Et7-mc-txq-2)#shape rate 300000
      switch(config-if-Et7-mc-txq-2)#exit
      switch(config-if-Et7)#show qos interface ethernet 7
      Ethernet7:
      
       Port shaping rate: 5000000Kbps
      
       Tx-Queue BandwidthShape Rate Priority Priority Group
      (percent) (Kbps)
       ----------------------------------------------------------------
      UC7 N/A disabledstrict1
      UC6 N/A disabledstrict1
      MC3 N/A disabledstrict1
      UC5 N/A 400000strict0
      UC4 N/A disabledstrict0
      MC2 N/A 300000strict0
      UC3 N/A 400000strict0
      UC2 N/A disabledstrict0
      MC1 N/A disabledstrict0
      UC1 N/A 500000strict0
      UC0 N/A 500000strict0
      MC0 N/A 300000strict0
      
      switch(config-if-Et7)#

Configuring Queue Priority

Trident and Tomahawk platform switch queues are categorized into two priority groups. Priority group 1 queues have priority over priority 0 queues. The following lists display the priority group queues in order from higher priority to lower priority.

  • Priority Group 1: UC7, UC6, MC3
  • Priority Group 0: UC5, UC4, MC2, UC3, UC2, MC1, UC1, UC0, MC0

The priority (Trident and Tomahawk) command configures a transmit queue’s priority type:

  • The priority strict command configures the queue as a strict priority queue.
  • The no priority command configures the queue as a round robin queue.

A queue’s configuration as round robin also applies to all lower priority queues regardless of other configuration statements.

The bandwidth percent (Trident and Tomahawk) command configures a round robin queue’s bandwidth share. The cumulative operational bandwidth of all round robin queues is always 100%. If the cumulative configured bandwidth is greater than 100%, each port’s operational bandwidth is its configured bandwidth divided by the cumulative configured bandwidth.

Priority Group 1 queues (UC7, UC6, MC3) are not configurable as round robin queues. The bandwidth percent command is not available for these queues.

Examples
  • These commands configure unicast transmit queue 3 as a round robin queue, then allocates 5%, 15%, 25%, 35%, 8%, and 12% bandwidth to unicast transmit queues 0 through 3 and multicast transmit queues 0 and 1, respectively.
    The no priority statement for queue 3 also configures priority for all lower priority queues. Removing the statement reverts the other queues to strict priority type unless running-config contains a no priority statement for one of these queues.
    switch(config)#interface ethernet 7
    switch(config-if-Et7)#uc-tx-queue 3
    switch(config-if-Et7-uc-txq-3)#no priority
    switch(config-if-Et7-uc-txq-3)#bandwidth percent 5
    switch(config-if-Et7-uc-txq-3)#uc-tx-queue 2
    switch(config-if-Et7-uc-txq-2)#bandwidth percent 15
    switch(config-if-Et7-uc-txq-2)#uc-tx-queue 1
    switch(config-if-Et7-uc-txq-1)#bandwidth percent 25
    switch(config-if-Et7-uc-txq-1)#uc-tx-queue 0
    switch(config-if-Et7-uc-txq-0)#bandwidth percent 35
    switch(config-if-Et7-uc-txq-0)#mc-tx-queue 1
    switch(config-if-Et7-mc-txq-1)#bandwidth percent 12
    switch(config-if-Et7-mc-txq-1)#mc-tx-queue 0
    switch(config-if-Et7-mc-txq-0)#bandwidth percent 8
    switch(config-if-Et7-mc-txq-0)#show qos interface ethernet 7
    Ethernet7:
    
     Port shaping rate: disabled
    
     Tx-Queue BandwidthShape Rate Priority Priority Group
    (percent) (Kbps)
     ----------------------------------------------------------------
    UC7 N/A disabledstrict1
    UC6 N/A disabledstrict1
    MC3 N/A disabledstrict1
    UC5 N/A disabledstrict0
    UC4 N/A disabledstrict0
    MC2 N/A disabledstrict0
    UC3 5 disabled round-robin0
    UC215 disabled round-robin0
    MC112 disabled round-robin0
    UC125 disabled round-robin0
    UC035 disabled round-robin0
    MC0 8 disabled round-robin0
    
    switch(config-if-Et7-mc-txq-0)#
  • Changing the bandwidth percentage for unicast queue 3 to 30 changes the operational bandwidth of each queue to its configured bandwidth divided by 125% (8%+12%+30%+15%+25%+35%).
    switch(config-if-Et7-uc-txq-0)#uc-tx-queue 3
    switch(config-if-Et7-uc-txq-3)#bandwidth percent 30
    switch(config-if-Et7-uc-txq-3)#show qos interface ethernet 7
    Ethernet7:
    
     Port shaping rate: disabled
    
     Tx-Queue BandwidthShape Rate Priority Priority Group
    (percent) (Kbps)
     ----------------------------------------------------------------
    UC7 N/A disabledstrict1
    UC6 N/A disabledstrict1
    MC3 N/A disabledstrict1
    UC5 N/A disabledstrict0
    UC4 N/A disabledstrict0
    MC2 N/A disabledstrict0
    UC324 disabled round-robin0
    UC212 disabled round-robin0
    MC1 9 disabled round-robin0
    UC120 disabled round-robin0
    UC028 disabled round-robin0
    MC0 6 disabled round-robin0
    
    switch(config-if-Et7-uc-txq-3)#

ECN Configuration – Trident and Tomahawk Platform Switches

Explicit Congestion Notification (ECN)describes Explicit Congestion Notification (ECN).

ECN is independently configurable on all egress queues of each Ethernet interface. ECN settings for Port-Channels are applied on each of the channel’s member Ethernet interfaces. ECN is also globally configurable to mark packets from the shared pool used for dynamically allocating memory to the queues. Multicast packets contribute to the globally shared pool and can contribute to global level congestion that result in ECN marking of unicast packets queued after the multicast packets.

Average queue length is tracked for transmit queues and the global pool independently. When either entity reaches its maximum threshold, all subsequent packets are marked.

Although the switch does not limit the number of queues that can be configured for ECN, hardware table limitations restrict the number of queues (including the global shared pool) that can simultaneously implement ECN.

Theqos random-detect ecn global-buffer (Trident and Tomahawk) command enables ECN marking for globally shared packet memory and specifies minimum and maximum queue threshold sizes.

Examples
  • This command enables ECN marking of unicast packets from the global data pool and sets the minimum and maximum thresholds at 20 and 500 segments.
    switch(config)#qos random-detect ecn global-buffer minimum-threshold 20 segments 
    maximum-threshold 500 segments
    switch(config)#
  • This command disables ECN marking of unicast packets from the global data pool.
    switch(config)#no qos random-detect ecn global-buffer
    switch(config)#

    The random-detect ecn (Trident and Tomahawk) command enables ECN marking for the configuration mode unicast transmit queue and specifies threshold queue sizes.

  • These commands enable ECN marking of unicast packets from transmit queue 4 of Ethernet interface 15, setting thresholds at 10 and 100 segments.
    switch(config)#interface ethernet 15
    switch(config-if-Et15)#uc-tx-queue 4
    switch(config-if-Et15-uc-txq-4)#random-detect ecn minimum-threshold 10 segments 
    maximum-threshold 100 segments
    switch(config-if-Et15-uc-txq-4)#show active
    interface Ethernet15
     uc-tx-queue 4
    random-detect ecn minimum-threshold 10 segments maximum-threshold 100 
    segments
    switch(config-if-Et15-uc-txq-4)#exit
    switch(config-if-Et15)#
  • This command disables ECN marking of unicast packets from transmit queue 4 of Ethernet interface 15.
    switch(config-if-Et15-uc-txq-4)#no random-detect ecn
    switch(config-if-Et15-uc-txq-4)#show active
    interface Ethernet15
    switch(config-if-Et15-uc-txq-4)#exit
    switch(config-if-Et15)#

QoS Configuration: Trident II and Helix Platform Switches

CoS and DSCP Port Settings – Trident II and Helix Platform Switches

Configuring Port Trust Settings

The qos trust command configures the QoS port trust mode for the configuration mode interface. Trust enabled ports use packet CoS or DSCP values to classify traffic. The port-trust default for switched ports is cos. The port-trust default for routed ports is dscp.
  • qos trust cos specifies cos as the port’s port-trust mode.
  • qos trust dscp specifies dscp as the port’s port-trust mode.
  • no qos trust specifies untrusted as the port’s port-trust mode.

The show qos interfaces trust command displays the trust mode of specified interfaces.

Example

These commands configure and display the following trust modes:
  • Ethernet 7/1: dscp.
  • Ethernet 7/2: untrusted.
  • Ethernet 7/3: cos.
  • Ethernet 7/4: default as a switched port.
  • Ethernet 8/1: default as a routed port.
  • switch(config)#interface ethernet 7/1
    switch(config-if-Et7/1)#qos trust dscp
    switch(config-if-Et7/1)#interface ethernet 7/2
    switch(config-if-Et7/2)#no qos trust
    switch(config-if-Et7/2)#interface ethernet 7/3
    switch(config-if-Et7/3)#qos trust cos
    switch(config-if-Et7/3)#interface ethernet 7/4
    switch(config-if-Et7/4)#switchport
    switch(config-if-Et7/4)#default qos trust
    switch(config-if-Et7/4)#interface ethernet 8/1
    switch(config-if-Et8/1)#no switchport
    switch(config-if-Et8/1)#default qos trust
    switch(config-if-Et8/1)#show qos interface ethernet 7/1 - 8/1 trust
    
    Port Trust Mode
     Operational Configured
    ---------------------------------------------------------------
    Ethernet7/1DSCPDSCP
    Ethernet7/2UNTRUSTED UNTRUSTED
    Ethernet7/3COS COS
    Ethernet7/4COS DEFAULT
    Ethernet8/1DSCPDEFAULT
    
    switch(config-if-Et8/1)#

Configuring Default Port Settings

Default CoS and DSCP settings are assigned to individual port channel and Ethernet interfaces. These configuration mode interface commands specify the port’s default CoS and DSCP values.
  • qos cos configures a port’s default CoS value.
  • qos dscp configures a port’s default DSCP value.

Example

These commands configure default CoS (4) and DSCP (44) values on Ethernet interface 7/3.
switch(config)#interface ethernet 7/3
switch(config-if-Et7/3)#qos cos 4
switch(config-if-Et7/3)#qos dscp 44
switch(config-if-Et7/3)#show active

interface Ethernet7/3
 qos cos 4
 qos dscp 44

switch(config-if-Et7/3)#show qos interfaces ethernet 7/3

Ethernet7/3:
 Trust Mode: COS
 Default COS: 4
 Default DSCP: 44

switch(config-if-Et7/3)#

Traffic Class Derivations – Trident II and Helix Platform Switches

Traffic Classes describes traffic classes.

Note: Qos traffic policy is supported on Trident II platform switches.

Traffic Class Derivation Source

The following table displays the source for deriving a data stream’s traffic class.

Table 29. Traffic Class Derivation Source: Trident II Platform Switches
  Untrusted CoS Trusted DSCP Trusted
Untagged Non-IP Default CoS (port) Default CoS (port) Default DSCP (port)
Untagged IP Default CoS (port) Default CoS (port) DSCP (packet)
Tagged Non-IP Default CoS (port) CoS (packet) Default DSCP (port)
Tagged IP Default CoS (port) CoS (packet) DSCP (packet)

CoS and DSCP Port Settings – Trident II and Helix Platform Switchesdescribes the default CoS and DSCP settings for each port.

Mapping CoS to Traffic Class

The qos map coscommand assigns a traffic class to a list of CoS settings. Multiple commands create a complete CoS to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s CoS field or the port upon which it is received.

Example

This command assigns the traffic class of 5 to the classes of service 1, 3, 5, and 7.
switch(config)#qos map cos 1 3 5 7 to traffic-class 5
switch(config)#show qos maps
 Number of Traffic Classes supported: 8

 Cos-tc map:
 cos:01234567
 ----------------------------
 tc: 15254565

switch(config)#

The following table displays the default CoS–traffic class map on Trident II platform switches.

Table 30. Default CoS to Traffic Class Map: Trident II Platform Switches
Inbound CoS 0 1 2 3 4 5 6 7
Traffic Class 1 0 2 3 4 5 6 7

Mapping DSCP to Traffic Class

The qos map dscp command assigns a traffic class to a set of DSCP values. Multiple commands create a complete DSCP to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s DSCP field or the chip upon which it is received.

Example

This command assigns the traffic class of 0 to DSCP values of 12, 24, 41, and 44-47.
switch(config)#qos map dscp 12 24 41 44 45 46 47 to traffic-class 0
switch(config)#show qos maps 
 Number of Traffic Classes supported: 8

 Dscp-tc map:
 d1 :d2 0123456789
 --------------------------------------
0 : 1111111100
1 : 0000002222
2 : 2222033333
3 : 3344444444
4 : 5055000066
5 : 6666667777
6 : 7777

switch(config)#

The following table displays the default DSCP–traffic class map on Trident II platform switches.

Table 31. Default DSCP to Traffic Class Map: Trident II Platform Switches
Inbound DSCP 0-7 8-15 16-23 24-31 32-39 40-47 48-55 56-63
Traffic Class 1 0 2 3 4 5 6 7

CoS and DSCP Rewrite – Trident II and Helix Platform Switches

Rewriting CoS and DSCP describes the CoS and DSCP rewrite functions.

Traffic Class to CoS Rewrite Map

The CoS rewrite value is configurable and based on a data stream’s traffic class, as specified by the traffic class-CoS rewrite map. The qos map traffic-class to cos command assigns a CoS rewrite value to a list of traffic classes. Multiple commands create the complete traffic class–CoS rewrite map.

Example

This command assigns the CoS of two to traffic classes 1, 3, and 5.
switch(config)#qos map traffic-class 1 3 5 to cos 2
switch(config)#show qos map

 Number of Traffic Classes supported: 8

 Tc-cos map:
 tc: 01234567
 ----------------------------
 cos:12224267

switch(config)#

The following tabledisplays the default Traffic Class to CoS rewrite value map on Trident II platform switches.

Table 32. Default Traffic Class to CoS Rewrite Value Map: Trident II Platform Switches

Traffic Class

0

1

2

3

4

5

6

7

CoS Rewrite Value

1

0

2

3

4

5

6

7

Traffic Class to DSCP Rewrite Map

The DSCP rewrite value is configurable and based on a data stream’s traffic class, as specified by the traffic class-DSCP rewrite map. The qos map traffic-class to dscp command assigns a DSCP rewrite value to a list of traffic classes. Multiple commands create the complete traffic class-DSCP rewrite map.

Example

This command assigns the DSCP value of 29 to traffic classes 2, 4, and 6.
switch(config)#qos map traffic-class 2 4 6 to dscp 29
switch(config)#show qos map

 Number of Traffic Classes supported: 8

 Tc-dscp map:
 tc:01234567
 -----------------------------
 dscp:80 29 24 29 40 29 56

switch(config)#

The following table displays the default traffic class to DSCP rewrite map on Trident II platform switches.

Table 33. Traffic Class to DSCP Rewrite Value Map: Trident II Platform Switches

Traffic Class

0

1

2

3

4

5

6

7

DSCP

8

0

16

24

32

40

48

56

Transmit Queues and Port Shaping – Trident II and Helix Platform Switches

Transmit Queues and Port Shaping describes transmit queues and port shaping.

A data stream’s traffic class determines the transmit queue it uses. The switch defines a single traffic class-transmit queue map for all Ethernet interfaces and is used for unicast and multicast traffic. The traffic class to transmit queue maps are configured globally and apply to all Ethernet and port channel interfaces. Theshow qos mapscommand displays the traffic class to transmit queue map.

Trident II platform switches have eight unicast (UC0 – UC7) and eight multicast (MC0 – MC7) queues. Each UCx-MCx queue set is combined into a single queue group (L1.x), which is exposed to the CLI through this command.

The following table displays the default traffic class to transmit queue maps.

Table 34. Default Traffic Class to Transmit Queue Map: Trident II Platform Switches
Traffic Class 0 1 2 3 4 5 6 7
Transmit Queue Group 0 1 2 3 4 5 6 7

Mapping Traffic Classes to a Transmit Queue

The qos map traffic-class to tx-queuecommand assigns traffic classes to a transmit queue. Multiple commands create the complete map.

Example

These commands assign traffic classes of 1, 3, and 5 to transmit queue 1, traffic classes 2, 4, and 6 to transmit queue 2, and traffic class 0 to transmit queue 0, then display the resultant map.
switch(config)#qos map traffic-class 1 3 5 to tx-queue 1
switch(config)#qos map traffic-class 2 4 6 to tx-queue 2
switch(config)#qos map traffic-class 0 to tx-queue 0
switch(config)#show qos maps
   Number of Traffic Classes supported: 8
 Number of Transmit Queues supported: 8

 Tc - tx-queue map:
 tc:01234567
 ---------------------------------
 tx-queue:01212127

switch(config)#

Entering a Transmit Queue Configuration Mode

Transmit queues are configurable on Ethernet ports and port channels. Queue parameters are configured in tx-queue configuration command mode, which is entered from the appropriate interface configuration mode. The tx-queue (Trident II)command places the switch in tx-queue configuration mode. The show qos interfaces displays the transmit queue configuration for a specified port.

Example

This command enters tx-queue configuration mode for transmit queue 3 of Ethernet interface 5.
switch(config)#interface ethernet 5
switch(config-if-Et5)#tx-queue 3
switch(config-if-Et5-txq-3)#

Configuring the Shape Rate – Port and Transmit Queues

A port’s shape rate specifies the port’s maximum outbound traffic bandwidth. A shape rate can also be configured for all transmit queues on each port. All shape rate commands use kbps to specify data rates.

Example

These commands configure a shape rate of 5 Gbs on Ethernet port 3, then configure the shape rate for the following transmit queues:
    • transmit queues 0, 1, and 2: 500 Mbps
    • transmit queues 3, 4, and 5: 400 Mbps

      switch(config)#interface ethernet 17/3
      switch(config-if-Et17/3)#shape rate 5000000
      switch(config-if-Et17/3)#tx-queue 0
      switch(config-if-Et17/3-txq-0)#shape rate 500000
      switch(config-if-Et17/3-txq-0)#tx-queue 1
      switch(config-if-Et17/3-txq-1)#shape rate 500000
      switch(config-if-Et17/3-txq-1)#tx-queue 3
      switch(config-if-Et17/3-txq-3)#shape rate 400000
      switch(config-if-Et17/3-txq-3)#tx-queue 4
      switch(config-if-Et17/3-txq-4)#shape rate 400000
      switch(config-if-Et17/3-txq-4)#tx-queue 5
      switch(config-if-Et17/3-txq-5)#shape rate 400000
      switch(config-if-Et17/3-txq-5)#exit
      switch(config-if-Et17/3)#show qos interface ethernet 17/3
      Ethernet17/3:
      
      Tx Bandwidth Shape RatePriority
       Queue Guaranteed (units) (units)
       ------------------------------------------------------------
       7- / -(-) - / -(-)SP / SP
       6- / -(-) - / -(-)SP / SP
       5- / -(-) 400 / 400( Mbps ) SP / SP
       4- / -(-) 400 / 400( Mbps ) SP / SP
       3- / -(-) 400 / 400( Mbps ) SP / SP
       2- / -(-) - / -(-)SP / SP
       1- / -(-) 500 / 500( Mbps ) SP / SP
       0- / -(-) 500 / 500( Mbps ) SP / SP
      
      switch(config-if-Et17/3)#

Configuring Queue Priority

Queue priority rank is denoted by the queue number; transmit queues with higher numbers have higher priority. Trident II supports strict priority queues; round robin queues are not supported.

The bandwidth guaranteed (Trident II) command configures specifies the minimum bandwidth for outbound traffic on the transmit queue.

Example

These commands configure a minimum egress bandwidth of 1 Mbps for transmit queue 4 of Ethernet interface 17/3.
switch(config-if-Et17/3)#tx-queue 4
switch(config-if-Et17/3-txq-4)#show qos interface ethernet 17/3

Tx Bandwidth Shape RatePriority
 Queue Guaranteed (units) (units)
 ------------------------------------------------------------
 7- / -(-) - / -(-)SP / SP
 6- / -(-) - / -(-)SP / SP
 5- / -(-) 400 / 400( Mbps ) SP / SP
 41 / 1( Mbps )400 / 400( Mbps ) SP / SP
 3- / -(-) 400 / 400( Mbps ) SP / SP
 2- / -(-) - / -(-)SP / SP
 1- / -(-) 500 / 500( Mbps ) SP / SP
 0- / -(-) 500 / 500( Mbps ) SP / SP

switch(config-if-Et17/3-txq-4)#

Ingress Policing on LAG

Ingress policing on a port-channel polices the matched traffic from all member interfaces combined, i.e. it provides aggregate policing and statistics (DCS-7050X, DCS-7010T, DCS-7250X, and DCS-7300X series). When a per-interface policer is attached to a port-channel, one set of TCAM entries is created for all member interfaces. The associated interface bitmap is updated, and aggregate policing is performed on all member interfaces.

Examples
  • These commands configure a service-policy (with policer action) on LAG by creating the service-policy and applying the service-policy on a port-channel.
    switch(config)#policy-map policy-1
    switch(config-pmap-qos-policy-1)#class class-1
    switch(config-pmap-c-qos-policy-1-class-1)#police cir 512000 bps bc 96000
    switch(config-pmap-c-qos-policy-1-class-1)#exit
    switch(config-pmap-qos-policy-1)#exit
    switch(config)#interface Et 4 / 5 / 4
    switch(config-if-Et4/5/4)#channel-group 2 mode active
    switch(config-if-Et4/5/4)#exit
    switch(config)#interface po2
    switch(config-if-Po2)#service-policy type qos input policy-1
    switch(config-if-Po2)#exit
    switch(config)#
  • These commands configure ACL policing in single-rate, two-color mode.
    switch(config)#class-map type qos match-any class1
    switch(config-cmap-qos-class1)#match ip access-group acl1
    switch(config-cmap-qos-class1)#exit
    switch(config)#policy-map type quality-of-service policy1
    switch(config-pmap-qos-policy1)#class class1
    switch(config-pmap-c-qos-policy1-class1)#police cir 512000 bc 96000
    switch(config-pmap-c-qos-policy1-class1)#exit
    switch(config-pmap-qos-policy1)#exit
    switch(config)#show policy-map
    Service-policy policy1
    
    Class-map: class1 (match-any)
    Match: ip access-group name acl1
     police rate 512000 bps burst-size 96000 bytes
    
    Class-map: class-default (match-any)
    
    switch(config)#

Fabric QoS -- – Trident II Platform Switches

EOS is optimized to support QoS configuration on the Fabric interfaces on 7250x and 7300 series switches. Configuring QoS on the Fabric interfaces in addition to front panel ports allows user to have end-to-end control and helps to manage traffic better over these switches. By default, tx queues are configured as strict priority on 7250X and 7300X series.

The following QoS configuration options are supported on Fabric interfaces on 7250x and 7300 series switches.

  • Guaranteed Bandwidth: In order to prevent queue starvation on fabric ports EOS supports minimum bandwidth configuration on per queue basis across all fabric ports.
  • Explicit Congestion Notification (ECN): EOS supports enabling ECN on a per queue basis across all fabric ports.
  • Priority Flow Control (PFC): Queue back-pressure propagates across the backplane such that flow control messages can be generated to the upstream devices. This is done by enabling PFC for the desired backplane traffic-classes.
  • Weight Round Robin (WRR): EOS supports configuring Weighted Round Robin (WRR) on a per queue basis across all fabric ports.

Configuring Fabric QoS on 7250X and 7300X Series

Fabric QoS is configured using a QoS profiles which is then applied on fabric interfaces on 7250x and 7300x series. Following are the steps to configure Fabric QoS.

  1. Use qos profile command to create a QoS profile.
  2. Use tx-queue (Trident II) command to configure a transmit queue on the configuration mode interface.
  3. Use bandwidth guaranteed (Trident II) command to specifie the minimum bandwidth for outbound traffic on the transmit queue.
  4. Use random-detect ecn (Trident) command to enable the ECN marking for the configuration mode unicast transmit queue and specifies threshold queue sizes.
  5. Use priority-flow-control priority command to configure the packet resolution setting on the configuration mode interface.
  6. Use interface fabric command to configure Fabric interface.
  7. Use service-profile command to apply the QoS profile to the Fabric interface.

    Examples

    • These commands create a QoS profile named fabricProfile with tx queue, bandwidth, ECN, PFC and DLB values defined in it and then the profile is attached to Fabric interface of the switch.

      Note: To support PFC on a particular priority, DLB is disabled for that priority.
      switch(config)#qos profile fabricProfile
      switch(config-qos-profile-fabricProfile)#tx-queue 0
      switch(config-qos-profile-fabricProfile-txq-0)#bandwidth guaranteed 10000 kbps
      switch(config-qos-profile-fabricProfile-txq-0)#random-detect ecn 
      minimum-threshold 10 mbytes maximum-threshold 10 mbytes
      switch(config-qos-profile-fabricProfile)#priority-flow-control priority 1 
      no-drop
      switch(config-qos-profile-fabricProfile)#priority-flow-control priority 6 
      no-drop dlb
    • Applying the QoS profile on ‘interface fabric’ of the switch.
      switch#configure terminal
      switch(config)#interface fabric
      switch(config-if-fabric)#service-profile fabricProfile

Displaying Fabric QoS Information

These show commands display the configured Fabric QoS information on the switch.
  • show qos profile [profile Name]: Displays the list of QoS profiles configured on the switch.
  • show qos interfaces fabric: Displays the profile applied on the fabric interface on the switch.

Examples

  • This command displays the fabricProfile profile information.
    switch#show qos profile fabricProfile
    qos profile fabricProfile
     priority-flow-control priority 1 no-drop
     priority-flow-control priority 6 no-drop dlb
     tx-queue 0
     bandwidth guaranteed 10000 kbps
     random-detect ecn minimum-threshold 10 mbytes maximum-threshold 10 mbytes
  • This command displays the profile applied on the fabric interface.
    switch#show qos interfaces fabric
     qos profile fabricProfile

ACL based QoS Configuration

ACL Based QoS (DCS-7160)

The IPv4 ACL based QoS is enabled on switches through policy-map configuration. The ACL based QoS can be configured on front panel ports, port-channel interfaces on DCS-7160 series switches.

ACL based QoS on SVIs

The ACL based QoS policy applied on SVIs modify the QoS parameters for SVI traffic (L3 VLAN) based on ACL classification. The ACL based QoS on Switched Virtual Interface (SVI) ports is supported on DCS-7500E, DCS-7280E, DCS-7010, DCS-7050, DCS-7050X, DCS-7250X, DCS-7300X, DCS-7020TR.

 

ACL Sharing on QoS

The ACLs applied on QoS shares the hardware resources (TCAM) when potentially large QoS policy-maps are applied to multiple SVIs. For ACL based QoS on SVIs in sharing mode we share TCAM for class-maps without policer action and replicate entries for policer class-maps. The following platforms support ACL sharing on Qos - HelixFour, TridentTwo, Tomahawk, TomahawkPlus, TomahawkTwo, Trident3, TridentThreeXThree.

The QoS actions is applicable only to the routed traffic flowing through the members of the corresponding VLAN.

The steps to configure ACL based QoS is as follows:

  1. Create a access list using ip access-list command.
  2. Create a class map and attach it to the access list using class-map command.
  3. Create a policy and attach the class map to the policy created, using the policy-map command.
  4. Apply the policy to the interface using the service-policy input command.

    Examples

    • These command configure the access list acl1.

      switch(config)#ip access-list acl1
      switch(config-acl-acl1)#permit ip 10.1.1.1/24 any
      switch(config-acl-acl1)#exit
      
      
    • These commands configure the class map class1.

      switch(config)#class-map match-any class1
      switch(config-cmap-qos-class1)#match ip access-group acl1
      switch(config-cmap-qos-class1)#exit
      
      
    • These commands configure the policy map policy1.

      switch(config)#policy-map policy1 
      switch(config-pmap-qos-policy1)#class class1
      switch(config-pmap-c-qos-policy1-class1)#set dscp 20
      switch(config-pmap-c-qos-policy1-class1)#set traffic-class 2
      switch(config-pmap-c-qos-policy1-class1)#exit 
      switch(config-pmap-c-qos-policy1)#exit
      
      
    • These commands apply the policy1 to the interface Ethernet 1/1.

      switch(config)#interface Et1/1
      switch(config-if-Et1/1)#service-policy input policy1 
      switch(config-if-Et1/1)#exit
      
      
    • These commands configure a ACL based QoS on the SVI interface VLAN10.

      switch(config)#interface vlan 10
      switch(config-if-Vl10)#service-policy [type qos] input policy1
      
      
    • This command enables the resource (hardware) sharing when a ACL based QoS is attached to VLAN interface. The no form of the command disables it.

      switch(config)#hardware access-list qos resource sharing vlan in  

    Show Commands

    The following show commands display the status of policy-maps programmed on the interface, for more information on these commands refer Quality of Service Configuration Commands.

    • show policy-map [policy-name]: Displays the policy-map programming status.
    • show policy-map interface interface id: Displays the policy-map that is currently programmed on the interface.
    • show policy-map [policy-name] counters: Displays the policy-map traffic hits.
    • show platform xp qos tcam [hits]: Displays the TCAM entries programmed for each policy-map as well as the traffic hits. The hits option is used to see the TCAM entries with nonzero traffic hits.
    • show run | grep sharing: Displays if whether the ACL based QoS is enabled or disabled.
    • show platform trident tcam shared vlan interface-class-id: Displays what SVIs are currently sharing the QoS policy-map.
    • show platform trident tcam qos detail: Displays the list of all the SVIs that are sharing the TCAM entries.

Limitations

  • Maximum number of TCAM entries that can be programmed in hardware for all QoS policy-maps on the box is 1024.
  • Layer 4 port ranges are not supported for ACL based QoS. The ranges will be expanded into multiple TCAM rules and programmed in the hardware.
  • Configured policer rate should be above 1mbps and recommended burst value is 2000 bytes.
  • Policer action can’t be associated with policy-maps applied to Port-Channels.

The following are the limitations specific to DCS-7500E, DCS-7280E and DCS-7020TR:

  • User cannot apply more than 31 QoS service policies per chip on L3 interfaces.
    • When different QoS service policies are applied to the SVI and its member interfaces, the behavior is indeterministic.
    • When QoS service policies are applied on SVIs with partial failures due to limited hardware resources, any event that triggers a forwarding agent restart will lead to indeterministic behavior.
    • When QoS service policies are applied on 2 SVIs, any event that triggers the VLAN membership change of a member interface may result in a policy-map programming failure. To change the VLAN membership, remove the interface from the first VLAN and then add it to the other.
    • Outgoing COS rewrite is not supported.
    • QoS policy-map counters are not supported.

The following are the limitations specific to DCS-7010, DCS-7050, DCS-7050X, DCS-7250X, DCS-7300X:

  • TCAM resources won’t be shared for the same policy-map applied to multiple SVIs.
  • Policy-map applied to a SVI will result in TCAM allocation on all chips irrespective of whether the SVI members are present or not.

When QoS service policies are applied to both SVI and its member interfaces and packets hit both policies, the behavior is indeterministic.

Quality of Service Configuration Commands

QoS Data Field and Traffic Class Configuration Commands

QoS and ECN Display Commands

ECN Configuration Commands

Transmit Queue and Port Shaping Commands – Arad and Jericho Platforms

Transmit Queue and Port Shaping Commands – FM6000 Platform

Transmit Queue and Port Shaping Commands – Helix Platform

Transmit Queue and Port Shaping Commands – Petra Platform

Transmit Queue and Port Shaping Commands – Trident and Tomahawk Platform

Transmit Queue and Port Shaping Commands – Trident II Platform

bandwidth guaranteed (Helix)

The bandwidth guaranteed command specifies the minimum bandwidth for outbound traffic on the transmit queue. By default, no bandwidth is guaranteed to any transmit queue.

The no bandwidth guaranteed and default bandwidth guaranteed commands remove the minimum bandwidth guarantee on the transmit queue by deleting the corresponding bandwidth guaranteed command from running-config.

Command Mode

Tx-Queue Configuration

Command Syntax

bandwidth guaranteed rate DATA_MIN

no bandwidth guaranteed

default bandwidth guaranteed

Parameters
  • DATA_MINMinimum bandwidth. Value range varies with data unit:
    • <8 to 40000000>  8 to 40000000 kbytes per second.
    • <8 to 40000000>kbps  8 to 40000000 kbytes per second.
    • <1 to 60000000>pps  1 to 60000000 packets per second.

Related Command

tx-queue (Helix) places the switch in tx-queue configuration mode.

Example
These commands configure a minimum egress bandwidth of 1 Mbps for transmit queue 4 of Ethernet interface 17/3.
switch(config)#interface ethernet 17
switch(config-if-Et17)#tx-queue 4
switch(config-if-Et17-txq-4)#bandwidth guaranteed 1000 kbps
switch(config-if-Et17-txq-4)#show qos interfaces ethernet 17

Ethernet17/3:
 Trust Mode: COS
 Default COS: 0
 Default DSCP: 0

 Port shaping rate: disabled

Tx Bandwidth Shape RatePriority
 Queue Guaranteed (units) (units)
 ------------------------------------------------------------
 7- / -(-) - / -(-)SP / SP
 6- / -(-) - / -(-)SP / SP
 5- / -(-) - / -(-)SP / SP
 41 / 1( Mbps )- / -(-)SP / SP
 3- / -(-) - / -(-)SP / SP
 2- / -(-) - / -(-)SP / SP
 1- / -(-) - / -(-)SP / SP
 0- / -(-) - / -(-)SP / SP

Note: Values are displayed as Operational/Configured

switch(config-if-Et17-txq-4)#

bandwidth guaranteed (Trident II)

The bandwidth guaranteed command specifies the minimum bandwidth for outbound traffic on the transmit queue. By default, no bandwidth is guaranteed to any transmit queue.

The no bandwidth guaranteed and default bandwidth guaranteed commands remove the minimum bandwidth guarantee on the transmit queue by deleting the corresponding bandwidth guaranteed command from running-config.

Command Mode

Tx-Queue Configuration

Command Syntax

bandwidth guaranteed rate DATA_MIN

no bandwidth guaranteed

default bandwidth guaranteed

Parameters
  • DATA_MIN minimum bandwidth. Value range varies with data unit:
    • <8 to 40000000>      8 to 40000000 kbytes per second.
    • <8 to 40000000>kbps      8 to 40000000 kbytes per second.
    • <1 to 60000000>pps      1 to 60000000 packets per second.

Related Commands

tx-queue (Trident II) places the switch in tx-queue configuration mode.

Example
These commands configure a minimum egress bandwidth of 1 Mbps for transmit queue 4 of Ethernet interface 17/3.
switch(config)#interface ethernet 17/3
switch(config-if-Et17/3)#tx-queue 4
switch(config-if-Et17/3-txq-4)#bandwidth guaranteed 1000 kbps
switch(config-if-Et17/3-txq-4)#show qos interfaces ethernet 17/3

Ethernet17/3:
 Trust Mode: COS
 Default COS: 0
 Default DSCP: 0

 Port shaping rate: disabled

Tx Bandwidth Shape RatePriority
 Queue Guaranteed (units) (units)
 ------------------------------------------------------------
 7- / -(-) - / -(-)SP / SP
 6- / -(-) - / -(-)SP / SP
 5- / -(-) - / -(-)SP / SP
 41 / 1( Mbps )- / -(-)SP / SP
 3- / -(-) - / -(-)SP / SP
 2- / -(-) - / -(-)SP / SP
 1- / -(-) - / -(-)SP / SP
 0- / -(-) - / -(-)SP / SP

Note: Values are displayed as Operational/Configured

switch(config-if-Et17/3-txq-4)#

bandwidth percent (Arad/Jericho)

The bandwidth percent command configures the bandwidth share of the transmit queue when configured for round robin priority. Bandwidth is allocated to all queues based on the cumulative configured bandwidth of all the port’s round robin queues.

The cumulative operational bandwidth of all round robin queues is always less than or equal to 100%. If the cumulative configured bandwidth is greater than 100%, each port’s operational bandwidth is its configured bandwidth divided by the cumulative configured bandwidth.

The no bandwidth percent and default bandwidth percent commands restore the default bandwidth share of the transmit queue by removing the corresponding bandwidth percent command from running-config.

Command Mode

Tx-Queue Configuration

Command Syntax

bandwidth percent proportion

no bandwidth percent

default bandwidth percent

Parameters

proportion Bandwidth percentage assigned to queues. Values range from 1 to 100.

Related Command

tx-queue (Arad/Jericho) places the switch in tx-queue configuration mode.

Examples
  • These commands configure queues 0 through 3 (Ethernet interface 3/5/1) as round robin, then allocate bandwidth for three queues at 30% and one queue at 10%.
    switch(config)#interface ethernet 3/5/1
    switch(config-if-Et3/5/1)#tx-queue 3
    switch(config-if-Et3/5/1-txq-3)#no priority
    switch(config-if-Et3/5/1-txq-3)#bandwidth percent 10
    switch(config-if-Et3/5/1-txq-3)#tx-queue 2
    switch(config-if-Et3/5/1-txq-2)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-2)#tx-queue 1
    switch(config-if-Et3/5/1-txq-1)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-1)#tx-queue 0
    switch(config-if-Et3/5/1-txq-0)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-0)#show qos interfaces ethernet 3/5/1
    
    Ethernet3/5/1:
    
    TxBandwidth Shape RatePriorityECN
     Queue(percent)(units)
     -----------------------------------------------------
     7- / -- / -(-)SP / SPD
     6- / -- / -(-)SP / SPD
     5- / -- / -(-)SP / SPD
     4- / -- / -(-)SP / SPD
     3 10 / 10 - / -(-)RR / RRD
     2 30 / 30 - / -(-)RR / SPD
     1 30 / 30 - / -(-)RR / SPD
     0 30 / 30 - / -(-)RR / SPD
    
    switch(config-if-Et3/5/1-txq-0)#
  • These commands re-configure the bandwidth share of the fourth queue at 30%.
    switch(config-if-Et3/5/1-txq-0)#tx-queue 3
    switch(config-if-Et3/5/1-txq-3)#bandwidth percent 30
    switch(config-if-Et3/5/1-txq-3)#show qos interfaces ethernet 3/5/1
    
    Ethernet3/5/1:
    
     Port shaping rate: disabled
    
    TxBandwidth Shape RatePriorityECN
     Queue(percent)(units)
     -----------------------------------------------------
     7- / -- / -(-)SP / SPD
     6- / -- / -(-)SP / SPD
     5- / -- / -(-)SP / SPD
     4- / -- / -(-)SP / SPD
     3 24 / 30 - / -(-)RR / RRD
     2 24 / 30 - / -(-)RR / SPD
     1 24 / 30 - / -(-)RR / SPD
     0 24 / 30 - / -(-)RR / SPD
    
    Note: Values are displayed as Operational/Configured
    
    switch(config-if-Et3/5/1-txq-3)#
  • These commands configure the bandwidth share of the fourth queue at 2%.
    switch(config-if-Et3/5/1-txq-3)#bandwidth percent 2
    switch(config-if-Et3/5/1-txq-3)#show qos interfaces ethernet 3/5/1
    
    Ethernet3/5/1:
    
     Port shaping rate: disabled
    
    TxBandwidth Shape RatePriorityECN
     Queue(percent)(units)
     -----------------------------------------------------
     7- / -- / -(-)SP / SPD
     6- / -- / -(-)SP / SPD
     5- / -- / -(-)SP / SPD
     4- / -- / -(-)SP / SPD
     32 / 2- / -(-)RR / RRD
     2 30 / 30 - / -(-)RR / SPD
     1 30 / 30 - / -(-)RR / SPD
     0 30 / 30 - / -(-)RR / SPD
    
    Note: Values are displayed as Operational/Configured
    
    switch(config-if-Et3/5/1-txq-3)#

bandwidth percent (FM6000)

The bandwidth percent command configures the bandwidth share of the transmit queue when configured for round robin priority. Bandwidth is allocated to all queues based on the cumulative configured bandwidth of all the port’s round robin queues.

The cumulative operational bandwidth of all round robin queues is always less than or equal to 100%. If the cumulative configured bandwidth is greater than 100%, each port’s operational bandwidth is its configured bandwidth divided by the cumulative configured bandwidth.

The no bandwidth percent and default bandwidth percent commands restore the default bandwidth share of the transmit queue by removing the corresponding bandwidth percent command running-config.

Command Mode

Tx-Queue Configuration

Command Syntax

bandwidth percent proportion

no bandwidth percent

default bandwidth percent

Parameters

proportion     Configured bandwidth percentage. Value ranges from 1 to 100. Default value is 0.

Related Command

tx-queue (FM6000) places the switch in tx-queue configuration mode.

Examples
  • These commands configure queues 0 through 3 (Ethernet interface 19) as round robin, then allocate bandwidth for three queues at 30% and one queue at 10%.
    switch(config)#interface Ethernet 19
    switch(config-if-Et19)#tx-queue 3
    switch(config-if-Et19-txq-3)#no priority
    switch(config-if-Et19-txq-3)#bandwidth percent 10
    switch(config-if-Et19-txq-3)#tx-queue 2
    switch(config-if-Et19-txq-2)#bandwidth percent 30
    switch(config-if-Et19-txq-2)#tx-queue 1
    switch(config-if-Et19-txq-1)#bandwidth percent 30
    switch(config-if-Et19-txq-1)#tx-queue 0
    switch(config-if-Et19-txq-0)#bandwidth percent 30
    switch(config-if-Et19-txq-0)#show qos interface ethernet 19
    
    Ethernet19:
     Trust Mode: COS
     Default COS: 0
     Default DSCP: 0
    
     Port shaping rate: disabled
    
     TxBandwidthBandwidthShape Rate PriorityECN/WRED
     Queue (percent)Guaranteed (units)units)
     
    -----------------------------------------------------------------------
    7 - / -- / - ( - ) - / -( - )SP / SPD
    6 - / -- / - ( - ) - / -( - )SP / SPD
    5 - / -- / - ( - ) - / -( - )SP / SPD
    4 - / -- / - ( - ) - / -( - )SP / SPD
    310 / 10 - / - ( - ) - / -( - )RR / RRD
    230 / 30 - / - ( - ) - / -( - )RR / SPD
    130 / 30 - / - ( - ) - / -( - )RR / SPD
    030 / 30 - / - ( - ) - / -( - )RR / SPD
    
    Note: Values are displayed as Operational/Configured
    
    Legend:
    RR -> Round Robin
    SP -> Strict Priority
     --> Not Applicable / Not Configured
    ECN/WRED: L -> Queue Length ECN Enabled W -> WRED Enabled D -> Disabled
    
  • These commands re-configure the bandwidth share of transmit queue 3 at 30%.
    cp118.14:04:20#config
    cp118.14:04:23(config)#interface Ethernet 19
    cp118.14:04:47(config-if-Et19-txq-0)#tx-queue 3
    cp118.14:04:59(config-if-Et19-txq-3)#bandwidth percent 30
    cp118.14:05:16(config-if-Et19-txq-3)#show qos interface ethernet 19
    
    Ethernet19:
     Trust Mode: COS
     Default COS: 0
     Default DSCP: 0
    
     Port shaping rate: disabled
    
     Tx Bandwidth BandwidthShape RatePriorityECN/WRED
    Queue (percent) Guaranteed (units)(units)
     
    ------------------------------------------------------------------------
    7 - / -- / - ( - ) - / -( - ) SP / SPD
    6 - / -- / - ( - ) - / -( - ) SP / SPD
    5 - / -- / - ( - ) - / -( - ) SP / SPD
    4 - / -- / - ( - ) - / -( - ) SP / SPD
    324 / 30 - / - ( - ) - / -( - ) RR / RRD
    224 / 30 - / - ( - ) - / -( - ) RR / SPD
    124 / 30 - / - ( - ) - / -( - ) RR / SPD
    024 / 30 - / - ( - ) - / -( - ) RR / SPD
    
    Note: Values are displayed as Operational/Configured
    
    Legend:
    RR -> Round Robin
    SP -> Strict Priority
     --> Not Applicable / Not Configured
    ECN/WRED: L -> Queue Length ECN Enabled W -> WRED Enabled D -> Disabled
    
  • These commands re-configure the bandwidth share of transmit queue 3 at 2%.
    cp118.14:09:37(config-if-Et19-txq-3)#bandwidth percent 2
    cp118.14:12:56(config-if-Et19-txq-3)#show qos interface ethernet 19
    
    Ethernet19:
     Trust Mode: COS
     Default COS: 0
     Default DSCP: 0
    
     Port shaping rate: disabled
    
     Tx BandwidthBandwidth Shape RatePriorityECN/WRED
    Queue (percent)Guaranteed (units)(units)
     
    ----------------------------------------------------------------------
    7 - / -- / - ( - ) - / -( - ) SP / SP D
    6 - / -- / - ( - ) - / -( - ) SP / SP D
    5 - / -- / - ( - ) - / -( - ) SP / SP D
    4 - / -- / - ( - ) - / -( - ) SP / SP D
    3 2 / 2- / - ( - ) - / -( - ) RR / RR D
    230 / 30 - / - ( - ) - / -( - ) RR / SP D
    130 / 30 - / - ( - ) - / -( - ) RR / SP D
    030 / 30 - / - ( - ) - / -( - ) RR / SP D
    
    Note: Values are displayed as Operational/Configured
    
    Legend:
    RR -> Round Robin
    SP -> Strict Priority
     --> Not Applicable / Not Configured
    ECN/WRED: L -> Queue Length ECN Enabled W -> WRED Enabled D -> Disabled
    

bandwidth percent (Petra)

The bandwidth percent command configures the bandwidth share of the transmit queue when configured for round robin priority. Bandwidth is allocated to all queues based on the cumulative configured bandwidth of all the port’s round robin queues.

The cumulative operational bandwidth of all round robin queues is always less than or equal to 100%. If the cumulative configured bandwidth is greater than 100%, each port’s operational bandwidth is its configured bandwidth divided by the cumulative configured bandwidth.

The no bandwidth percent and default bandwidth percent commands restore the default bandwidth share of the transmit queue by removing the corresponding bandwidth percent command running-config.

Command Mode

Tx-Queue Configuration

Command Syntax

bandwidth percent proportion

no bandwidth percent

default bandwidth percent

Parameters

proportionBandwidth percentage assigned to queues. Values range from 1 to 100.

Related Command

tx-queue (Petra)places the switch in tx-queue configuration mode.

Examples
  • These commands configure queues 0 through 3 (Ethernet interface 3/28) as round robin, then allocate bandwidth for three queues at 30% and one queue at 10%.
    switch(config)#interface ethernet 3/28
    switch(config-if-Et3/28)#tx-queue 3
    switch(config-if-Et3/28-txq-3)#no priority
    switch(config-if-Et3/28-txq-3)#bandwidth percent 10
    switch(config-if-Et3/28-txq-3)#tx-queue 2
    switch(config-if-Et3/28-txq-2)#bandwidth percent 30
    switch(config-if-Et3/28-txq-2)#tx-queue 1
    switch(config-if-Et3/28-txq-1)#bandwidth percent 30
    switch(config-if-Et3/28-txq-1)#tx-queue 0
    switch(config-if-Et3/28-txq-0)#bandwidth percent 30
    switch(config-if-Et3/28-txq-0)#show qos interface ethernet 3/28
    
    Ethernet3/28:
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
     -----------------------------------------------
    7 N/A disabledstrict
    6 N/A disabledstrict
    5 N/A disabledstrict
    4 N/A disabledstrict
    310 disabled round-robin
    230 disabled round-robin
    130 disabled round-robin
    030 disabled round-robin
    
    switch(config-if-Et3/28-txq-0)#
  • These commands re-configure the bandwidth share of the fourth queue at 30%.
    switch(config-if-Et3/28-txq-0)#tx-queue 3
    switch(config-if-Et3/28-txq-3)#bandwidth percent 30
    switch(config-if-Et3/28-txq-3)#show qos interface ethernet 3/28
    
    Ethernet3/28:
     Trust Mode: COS
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
    -----------------------------------------------
    7 N/A disabledstrict
    6 N/A disabledstrict
    5 N/A disabledstrict
    4 N/A disabledstrict
    324 disabled round-robin
    224 disabled round-robin
    124 disabled round-robin
    024 disabled round-robin
    
    switch(config-if-Et3/28-txq-3)#
  • These commands configure the bandwidth share of the fourth queue at 2%.
    switch(config-if-Et3/28)#tx-queue 3
    switch(config-if-Et3/28-txq-3)#bandwidth percent 2
    switch(config-if-Et3/28-txq-3)#show qos interface ethernet 3/28
    
    Ethernet3/28:
     Trust Mode: COS
    
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
     -----------------------------------------------
    7 N/A disabledstrict
    6 N/A disabledstrict
    5 N/A disabledstrict
    4 N/A disabledstrict
    3 2 disabled round-robin
    230 disabled round-robin
    130 disabled round-robin
    030 disabled round-robin
    
    switch(config-if-Et3/28-txq-3)#

bandwidth percent (Trident and Tomahawk)

The bandwidth percent command configures the bandwidth share of the transmit queue when configured for round robin priority. Bandwidth is allocated to all queues based on the cumulative configured bandwidth of all the port’s round robin queues.

The cumulative operational bandwidth of all round robin queues is always less than or equal to 100%. If the cumulative configured bandwidth is greater than 100%, each port’s operational bandwidth is its configured bandwidth divided by the cumulative configured bandwidth.

The no bandwidth percent and default bandwidth percent commands restore the default bandwidth share of the transmit queue by removing the corresponding bandwidth percent command running-config.

Command Mode

Mc-Tx-Queue configuration

Uc-Tx-Queue configuration

Command Syntax

bandwidth percent proportion

no bandwidth percent

default bandwidth percent

Parameters

proportion Bandwidth percentage assigned to queues. Values range from 1 to 100.

Related Commands
  • mc-tx-queue places the switch in mc-tx-queue configuration mode.
  • uc-tx-queue places the switch in uc-tx-queue configuration mode.
Examples
  • These commands configure unicast transmit queue 3 (and all other queues of lower priority) as round robin, then allocate bandwidth for unicast transmit queues 1, 2, and 3 at 30% and multicast transmit queue 1 at 10%.
    switch(config)#interface ethernet 7
    switch(config-if-Et7)#uc-tx-queue 3
    switch(config-if-Et7-uc-txq-3)#no priority
    switch(config-if-Et7-uc-txq-3)#bandwidth percent 30
    switch(config-if-Et7-uc-txq-3)#uc-tx-queue 2
    switch(config-if-Et7-uc-txq-2)#bandwidth percent 30
    switch(config-if-Et7-uc-txq-2)#uc-tx-queue 1
    switch(config-if-Et7-uc-txq-1)#bandwidth percent 30
    switch(config-if-Et7-uc-txq-1)#mc-tx-queue 1
    switch(config-if-Et7-mc-txq-1)#bandwidth percent 10
    switch(config-if-Et7-mc-txq-1)#show qos interfaces ethernet 7
    
    Ethernet7:
     Trust Mode: COS
     Default COS: 0
     Default DSCP: 0
    
     Port shaping rate: disabled
    
     Tx-Queue BandwidthShape Rate Priority Priority Group
    (percent) (Kbps)
     ----------------------------------------------------------------
    UC7 N/A disabledstrict1
    UC6 N/A disabledstrict1
    MC3 N/A disabledstrict1
    UC5 N/A disabledstrict0
    UC4 N/A disabledstrict0
    MC2 N/A disabledstrict0
    UC330 disabled round-robin0
    UC230 disabled round-robin0
    MC110 disabled round-robin0
    UC130 disabled round-robin0
    UC0 0 disabled round-robin0
    MC0 0 disabled round-robin0
    
    switch(config-if-Et7-mc-txq-1)#
  • These commands re-configure the bandwidth share of unicast queue 3 at 55%.
    switch(config-if-Et7-mc-txq-1)#uc-tx-queue 3
    switch(config-if-Et7-uc-txq-3)#bandwidth percent 55
    switch(config-if-Et7-uc-txq-3)#show qos interface ethernet 7
    
    Ethernet7:
     Trust Mode: COS
     Default COS: 0
     Default DSCP: 0
    
     Port shaping rate: disabled
    
     Tx-Queue BandwidthShape Rate Priority Priority Group
    (percent) (Kbps)
     ----------------------------------------------------------------
    UC7 N/A disabledstrict1
    UC6 N/A disabledstrict1
    MC3 N/A disabledstrict1
    UC5 N/A disabledstrict0
    UC4 N/A disabledstrict0
    MC2 N/A disabledstrict0
    UC344 disabled round-robin0
    UC224 disabled round-robin0
    MC1 8 disabled round-robin0
    UC124 disabled round-robin0
    UC0 0 disabled round-robin0
    MC0 0 disabled round-robin0
    
    switch(config-if-Et7-uc-txq-3)#

hardware access-list qos resource sharing vlan in

The hardware access-list qos resource sharing vlan in command enables the ACL based QoS resources sharing on a VLAN interface.

The no hardware access-list qos resource sharing vlan in disables the ACL based QoS resources sharing on a VLAN interface. By default this function is disabled.

Command Mode

Global Configuration

Command Syntax

hardware access-list qos resource sharing vlan in

no hardware access-list qos resource sharing vlan in

Example
This commands enables the ACL based QoS resources sharing on a VLAN interface.
switch(config)#hardware access-list qos resource sharing vlan in

interface fabric (Trident II)

The interface fabric command places the switch in Fabric-interface configuration mode and allows the user to attach the QoS profile to the fabric interface of the switch.

Command Mode

Global Configuration

Command Syntax

interface fabric

Example
This command places the switch in Fabric-interface configuration mode.
switch(config)#interface fabric
switch(config-if-fabric)#

mc-tx-queue

The mc-tx-queue command places the switch in mc-tx-queue configuration mode to configure a multicast transmit queue on the configuration mode interface. Mc-tx-queue configuration mode is not a group change mode; running-config is changed immediately after commands are executed. The exit command does not affect the configuration.

Trident and Tomahawk switches have four multicast queues (MC0 – MC03) and eight unicast queues (UC0 – UC7), categorized into two priority groups. All queues are exposed through the CLI and are user configurable.
  • Priority Group 1: UC7, UC6, MC3
  • Priority Group 0: UC5, UC4, MC2, UC3, UC2, MC1, UC1, UC0, MC0

The exit command returns the switch to the configuration mode for the base Ethernet or port channel interface.

The no mc-tx-queue and default mc-tx-queue commands remove the configuration for the specified transmit queue by deleting the all corresponding mc-tx-queue mode commands from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

mc-tx-queue queue_level

Parameters
  • queue_level      The multicast transmit queue number. Values range from 0 to 3.

Related Command

: Configures unicast transmit queues on Trident and Tomahawk platform switches.

Example
This command enters mc-tx-queue configuration mode for multicast transmit queue 3 of Ethernet interface 5.
switch(config)#interface ethernet 5
switch(config-if-Et5)#mc-tx-queue 3
switch(config-if-Et5-mc-txq-3)#

platform petraA traffic-class

The platform petraA traffic-class command configures the default traffic class used by all ports on a specified chip. The default traffic class is implemented by Petra platform switches to replace qos cos and qos dscp commands. Traffic class values range from 0 to 7. The default traffic class is 1.

When platform ? returns Petra:
  • CoS trusted ports: inbound untagged packets are assigned to the default traffic class. Tagged packets are assigned to the traffic class that corresponds to the contents of its CoS field.
  • DSCP trusted ports: inbound non-IP packets are assigned to the default traffic class. IP packets are assigned to the traffic class that corresponds to the contents of its DSCP field.
  • Untrusted ports: all inbound packets are assigned to the default traffic class.

The no platform petraA traffic-class and default platform petraA traffic-class commands restore the default traffic class of one for all ports on the specified chips by deleting the corresponding platform petraA traffic-class command from running-config.

Command Mode

Global Configuration

Command Syntax

platform petraA [CHIP_NAME] traffic-class tc_value

no platform petraA [CHIP_NAME] traffic-class

default platform petraA [CHIP_NAME] traffic-class

Parameters
  • CHIP_NAME Trust mode assigned to the specified ports. Port designation options include:
    • <no parameter>     All ports on the switch.
    • module cardX     All ports on specified linecard (7500 Series).
    • petra cardX / chipY      All ports on PetraA chip chipY on linecard cardX (7500 Series).
    • petra -chipZ      All ports on PetraA chip chipZ (7048 Series)

7500 Series

Switches can contain up to eight linecards. CardX varies from 3 to 10.

Each linecard contains six PetraA chips. Each chip controls eight ports. ChipY varies from 0 to 5:
  • 0 controls ports 1 through 8:
    • 1 controls ports 9 through 16.
    • 2 controls ports 17 through 24.
    • 3 controls ports 25 through 32.
    • 4 controls ports 33 through 40.
    • 5 controls ports 41 through 48.

7048 Series

Each switch contains two PetraA chips. ChipZ varies from 0 to 1:
  • 0 controls ports 1 through 32.
    • 1 controls ports 33 through 52.
  • tc_value      Traffic class value. Values range from 0 to 7. Default value is 1.

Related Command

show platform petraA traffic-class displays the traffic class assignment on all specified Petra chips.

Example
This command configures the default traffic class 6 for ports 25-32 on linecard 5.
switch(config)#platform petraA petra5/3 traffic-class 6
switch(config)#

priority (Arad/Jericho)

The priority command specifies the priority of the transmit queue. The switch supports two queue priorities:
  • strict priority: contents are removed from the queue - subject to maximum bandwidth limits, before data from lower priority queues. The default setting on all queues is strict priority.
  • round robin priority: contents are removed proportionately from all round robin queues - subject to maximum bandwidth limits assigned to the strict priority queues.

Tx-queue 7 is set to strict priority and is not configurable.

When a queue is configured as a round robin queue, all lower priority queues also function as round robin queues. A queue’s numerical label denotes its priority: higher labels denote higher priority. Tx-queue 6 has higher priority than Tx-queue 5, and Tx-queue 0 has the lowest priority.

The priority strict and default priority commands configure a transmit queue to function as a strict priority queue unless a higher priority queue is configured as a round robin queue.

The no priority command configures a transmit queue as a round robin queue. All lower priority queues also function as round robin queues regardless of their configuration.

Command Mode

Tx-Queue Configuration

Command Syntax

priority strict

no priority

default priority

Related Command

tx-queue (Arad/Jericho) places the switch in tx-queue configuration mode.

Example
These commands perform the following on Ethernet interface 3/4/1:
  • Displays the default state of all transmit queues.
  • Configures transmit queue 3 as a round robin queue.
  • Displays the effect of the no priority command on all transmit queues on the interface.
switch(config)#interface ethernet 3/4/1
switch(config-if-Et3/4/1)#show qos interfaces ethernet 3/4/1

Ethernet3/4/1:

TxBandwidth Shape RatePriorityECN
 Queue(percent)(units)
 -----------------------------------------------------
 7- / -- / -(-)SP / SPD
 6- / -- / -(-)SP / SPD
 5- / -- / -(-)SP / SPD
 4- / -999 / 1000 ( Mbps ) SP / SPD
 3- / -999 / 1000 ( Mbps ) SP / SPD
 2- / -- / -(-)SP / SPD
 1- / -- / -(-)SP / SPD
 0- / -- / -(-)SP / SPD

Note: Values are displayed as Operational/Configured

switch(config-if-Et3/4/1)#tx-queue 3
switch(config-if-Et3/4/1-txq-3)#no priority
switch(config-if-Et3/4/1-txq-3)#show qos interfaces ethernet 3/4/1

Ethernet3/4/1:

TxBandwidth Shape RatePriorityECN
 Queue(percent)(units)
 -----------------------------------------------------
 7- / -- / -(-)SP / SPD
 6- / -- / -(-)SP / SPD
 5- / -- / -(-)SP / SPD
 4- / -999 / 1000 ( Mbps ) SP / SPD
 3 25 / -999 / 1000 ( Mbps ) RR / RRD
 2 25 / -- / -(-)RR / SPD
 1 25 / -- / -(-)RR / SPD
 0 25 / -- / -(-)RR / SPD

Note: Values are displayed as Operational/Configured

switch(config-if-Et3/4/1-txq-3)#

priority (FM6000)

The priority command specifies the priority of the transmit queue. The switch supports two queue priorities:
  • strict priority: contents are removed from the queue - subject to maximum bandwidth limits, before data from lower priority queues. The default setting on all queues is strict priority.
  • round robin priority: contents are removed proportionately from all round robin queues - subject to maximum bandwidth limits assigned to the strict priority queues.

When a queue is configured as a round robin queue, all lower priority queues also function as round robin queues. A queue’s numerical label denotes its priority: higher labels denote higher priority. Tx-queue 6 has higher priority than Tx-queue 5, and Tx-queue 0 has the lowest priority.

The priority strict and default priority commands configure a transmit queue to function as a strict priority queue unless a higher priority queue is configured as a round robin queue.

The no priority command configures a transmit queue as a round robin queue. All lower priority queues also function as round robin queues regardless of their configuration.

Command Mode

Tx-Queue Configuration

Command Syntax

priority strict

no priority

default priority

Related Command

tx-queue (FM6000) places the switch in tx-queue configuration mode.

Example
These commands perform the following on Ethernet interface 2:
  • Displays the default state of all transmit queues.
  • Configures transmit queue 3 as a round robin queue.
  • Displays the effect of the no priority command on all transmit queues on the interface.
    switch(config)#interface ethernet 19
    switch(config-if-Et19)#show qos interface ethernet 19
    
    Ethernet19:
     Trust Mode: COS
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
     -----------------------------------------------
    6 N/A disabledstrict
    5 N/A disabledstrict
    4 N/A disabledstrict
    3 N/A disabledstrict
    2 N/A disabledstrict
    1 N/A disabledstrict
    0 N/A disabledstrict
    
    switch(config-if-Et19)#tx-queue 3
    switch(config-if-Et19-txq-3)#no priority
    switch(config-if-Et19-txq-3)#show qos interface ethernet 19
    
    Ethernet19:
     Trust Mode: COS
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
     -----------------------------------------------
    6 N/A disabledstrict
    5 N/A disabledstrict
    4 N/A disabledstrict
    325 disabled round-robin
    225 disabled round-robin
    125 disabled round-robin
    025 disabled round-robin
    
    switch(config-if-Et19-txq-3)#

priority (Petra)

The priority command specifies the priority of the transmit queue. The switch supports two queue priorities:
  • strict priority: contents are removed from the queue - subject to maximum bandwidth limits, before data from lower priority queues. The default setting on all queues is strict priority.
  • round robin priority: contents are removed proportionately from all round robin queues - subject to maximum bandwidth limits assigned to the strict priority queues.

Tx-queue 7 is set to strict priority and is not configurable.

When a queue is configured as a round robin queue, all lower priority queues also function as round robin queues. A queue’s numerical label denotes its priority: higher labels denote higher priority. Tx-queue 6 has higher priority than Tx-queue 5, and Tx-queue 0 has the lowest priority.

The priority strict and default priority commands configure a transmit queue to function as a strict priority queue unless a higher priority queue is configured as a round robin queue.

The no priority command configures a transmit queue as a round robin queue. All lower priority queues also function as round robin queues regardless of their configuration.

Command Mode

Tx-Queue Configuration

Command Syntax

priority strict

no priority

default priority

Related Command

tx-queue (Petra) places the switch in tx-queue configuration mode.

Example
These commands perform the following on Ethernet interface 3/28:
  • Displays the default state of all transmit queues.
  • Configures transmit queue 3 as a round robin queue.
  • Displays the effect of the no priority command on all transmit queues on the interface.
    switch(config)#interface ethernet 3/28
    switch(config-if-Et3/28)#show qos interface ethernet 3/28
    
    Ethernet3/28:
     Trust Mode: COS
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
     -----------------------------------------------
    7 N/A disabledstrict
    6 N/A disabledstrict
    5 N/A disabledstrict
    4 N/A disabledstrict
    3 N/A disabledstrict
    2 N/A disabledstrict
    1 N/A disabledstrict
    0 N/A disabledstrict
    
    switch(config-if-Et3/28)#tx-queue 3
    switch(config-if-Et3/28-txq-3)#no priority
    switch(config-if-Et3/28-txq-3)#show qos interface ethernet 3/28
    
    Ethernet3/28:
     Trust Mode: COS
    
     Tx-Queue BandwidthShape Rate Priority
    (percent) (Kbps)
     -----------------------------------------------
    7 N/A disabledstrict
    6 N/A disabledstrict
    5 N/A disabledstrict
    4 N/A disabledstrict
    325 disabled round-robin
    225 disabled round-robin
    125 disabled round-robin
    025 disabled round-robin
    
    switch(config-if-Et3/28-txq-3)#

priority (Trident and Tomahawk)

The priority command specifies the priority of the transmit queue. The switch supports two queue priorities:
  • strict priority: contents are removed from the queue - subject to maximum bandwidth limits, before data from lower priority queues. The default setting on all other queues is strict priority.
  • round robin priority: contents are removed proportionately from all round robin queues - subject to maximum bandwidth limits assigned to the strict priority queues.
Trident and Tomahawk switches have eight unicast queues (UC0 – UC7) and four multicast queues (MC0 – MC03), categorized into two priority groups. Priority group 1 queues have priority over priority 0 queues. The following lists display the priority group queues in order from higher priority to lower priority.
  • Priority Group 1: UC7, UC6, MC3
  • Priority Group 0: UC5, UC4, MC2, UC3, UC2, MC1, UC1, UC0, MC0

Priority group 1 queues are strict priority queues and are not configurable as round robin. Priority 0 queues are strict priority by default and are configurable as round robin. When a queue is configured as a round robin queue, all lower priority queues automatically function as round robin queues.

The priority strict and default priority commands configure a transmit queue to function as a strict priority queue unless a higher priority queue is configured as a round robin queue.

The no priority command configures a transmit queue as a round robin queue. All lower priority queues also function as round robin queues regardless of their configuration.

Command Mode

Mc-Tx-Queue configuration

Uc-Tx-Queue configuration

Command Syntax

priority strict

no priority

default priority

Related Commands
  • mc-tx-queue places the switch in mc-tx-queue configuration mode.
  • uc-tx-queue places the switch in uc-tx-queue configuration mode.
Example
These commands perform the following on Ethernet interface 7:
  • Displays the default state of all transmit queues.
  • Configures transmit queue 3 as a round robin queue.
  • Displays the effect of the no priority command on all transmit queues on the interface.
    switch(config)#interface ethernet 7
    switch(config-if-Et7)#show qos interface ethernet 7
    
    Ethernet7:
     Trust Mode: COS
    
     Tx-Queue BandwidthShape Rate Priority Priority Group
    (percent) (Kbps)
     ----------------------------------------------------------------
    UC7 N/A disabledstrict1
    UC6 N/A disabledstrict1
    MC3 N/A disabledstrict1
    UC5 N/A disabledstrict0
    UC4 N/A disabledstrict0
    MC2 N/A disabledstrict0
    UC3 N/A disabledstrict0
    UC2 N/A disabledstrict0
    MC1 N/A disabledstrict0
    UC1 N/A disabledstrict0
    UC0 N/A disabledstrict0
    MC0 N/A disabledstrict0
    
    switch(config-if-Et7)#uc-tx-queue 3
    switch(config-if-Et7-uc-txq-3)#no priority
    switch(config-if-Et7-uc-txq-3)#show qos interface ethernet 7
    
    Ethernet7:
     Trust Mode: COS
    
     Tx-Queue BandwidthShape Rate Priority Priority Group
    (percent) (Kbps)
     ----------------------------------------------------------------
    UC7 N/A disabledstrict1
    UC6 N/A disabledstrict1
    MC3 N/A disabledstrict1
    UC5 N/A disabledstrict0
    UC4 N/A disabledstrict0
    MC2 N/A disabledstrict0
    UC320 disabled round-robin0
    UC216 disabled round-robin0
    MC116 disabled round-robin0
    UC116 disabled round-robin0
    UC016 disabled round-robin0
    MC016 disabled round-robin0
    
    switch(config-if-Et7-uc-txq-3)#

qos cos

The qos cos command specifies the default class of service (CoS) value of the configuration mode interface. CoS values range from 0 to 7. Default value is 0.

Arad, Jericho, fm6000, Trident, Tomahawk, and Trident II platform switches:
  • CoS trusted ports: the default CoS value determines the traffic class for inbound untagged packets. Tagged packets are assigned to the traffic class that corresponds to the contents of its CoS field.
  • Untrusted ports: the default CoS value determines the traffic class for all inbound packets.
Petra platform switches:
  • CoS trusted ports: inbound untagged packets are assigned to the default traffic class, as configured by platform petraA traffic-class. Tagged packets are assigned to the traffic class that corresponds to the contents of its CoS field.
  • Untrusted ports: all inbound packets are assigned to the default traffic class.

The no qos cos and default qos cos commands restore the port’s default CoS value to zero by deleting the corresponding qos cos command from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

qos cos cos_value

no qos cos

default qos cos

Parameters

cos_valueCoS value assigned to port. Value ranges from 0 to 7. Default value is 0.

Example
This command configures the default CoS 4 on Ethernet interface 8.
switch(config-if-Et8)#qos cos 4
switch(config-if-Et8)#

qos dscp

The qos dscp command specifies the default Differentiated Services Code Point (DSCP) value of the configuration mode interface. The default DSCP determines the traffic class for non-IP packets that are inbound on DSCP trusted ports. DSCP trusted ports determine the traffic class for inbound packets as follows:
  • Arad, Jericho, fm6000, Trident, Tomahawk, and Trident II platform switches:
    • non-IP packets: default DSCP value specified by qos dscp determines the traffic class.
    • IP packets: assigned to the traffic class corresponding to its DSCP field contents.
  • Petra platform switches:
    • non-IP packets: assigned to default traffic class configured by platform petraA traffic-class.
    • IP packets: assigned to the traffic class corresponding to its DSCP field contents.

The no qos dscp and default qos dscp commands restore the port’s default DSCP value to zero by deleting the corresponding qos dscp command from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

qos dscp dscp_value

no qos dscp

default qos dscp

Parameters

dscp_value DSCP value assigned to the port. Value ranges from 0 to 63. Default value is 0.

Example
This command sets the default DSCP of 44 on Ethernet 7 interface.
switch(config)#interface ethernet 7
switch(config-if-Et7)#qos dscp 44
switch(config-if-Et7)

qos map cos

The qos map cos command associates a traffic class to a list of class of service (CoS) settings. Multiple commands create a complete CoS to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s CoS field or the port upon which it is received.

The no qos map cos and default qos map cos commands restore the specified CoS values to their default traffic class setting by deleting the corresponding qos map cos statements from running-config.

Command Mode

Global Configuration

Command Syntax

qos map cos cos_value_1 [cos_value_2 ... cos_value_n] to traffic-class tc_value

no qos map cos cos_value_1 [cos_value_2 ... cos_value_n]

default qos map cos cos_value_1 [cos_value_2 ... cos_value_n]

Parameters
  • cos_value_x      Class of service (CoS) value. Value ranges from 0 to 7.
  • tc_value     Traffic class value. Value range varies by platform.

Default CoS to traffic class map varies by platform (Table 35).

Default Inbound CoS to Traffic Class Map

Table 35 displays the default CoS to traffic class map for each platform.

Table 35. Default CoS to Traffic Class Map
Inbound CoS untagged 0 1 2 3 4 5 6 7
Traffic Class (Arad /Jericho) Derived: use default CoS as inbound CoS 1 0 2 3 4 5 6 7
Traffic Class (FM6000) Derived: use default CoS as inbound CoS 1 0 2 3 4 5 6 7
Traffic Class (Helix) Derived: use default CoS as inbound CoS 1 0 2 3 4 5 6 7
Traffic Class (Petra) Assigned default traffic class 1 0 2 3 4 5 6 7
Traffic Class (Trident and Tomahawk) Derived: use default CoS as inbound CoS 1 0 2 3 4 5 6 7
Traffic Class (Trident II) Derived: use default CoS as inbound CoS 1 0 2 3 4 5 6 7
Related Commands
Example
This command assigns the traffic class 5 to the classes of service 1, 3, 5, and 7.
switch(config)#qos map cos 1 3 5 7 to traffic-class 5
switch(config)#

qos map dscp

The qos map dscp command associates a traffic class to a set of Differentiated Services Code Point (DSCP) values. Multiple commands create a complete DSCP to traffic class map. The switch uses this map to assign a traffic class to data packets on the basis of the packet’s DSCP field or the chip upon which it is received.

The no qos map dscp and default qos map dscp commands restore the specified DSCP values to their default traffic class settings by deleting corresponding qos map dscp statements from running-config.

Command Mode

Global Configuration

Command Syntax

qos map dscp dscpv_1 [dscpv_2...dscpv_n] to traffic-class tc_value

no qos map dscp dscpv_1 [dscpv_2...dscpv_n]

default qos map dscp dscpv_1 [dscpv_2...dscpv_n]

Parameters
  • dscpv_x      Differentiated services code point (DSCP) value. Value ranges from 0 to 63.
  • tc_value     Traffic class value. Value range varies by platform.

Default map varies by platform (Table 36).

Default Inbound DSCP to Traffic Class Map

Table 36 displays the default DSCP to traffic class map for each platform.

Table 36. Default DSCP to Traffic Class Map
Inbound DSCP 0-7 8-15 16-23 24-31 32-39 40-47 48-55 56-63
Traffic Class (Arad /Jericho) 1 0 2 3 4 5 6 7
Traffic Class (FM6000) 1 0 2 3 4 5 6 7
Traffic Class (Helix) 1 0 2 3 4 5 6 7
Traffic Class (Petra) 1 0 2 3 4 5 6 7
Traffic Class (Trident and Tomahawk) 1 0 2 3 4 5 6 7
Traffic Class (Trident II) 1 0 2 3 4 5 6 7
Example
This command assigns the traffic class 3 to the DSCP values of 12, 13, 25, and 37.
switch(config)#qos map dscp 12 13 25 37 to traffic-class 3
switch(config)#

qos map traffic-class to cos

The qos map traffic-class to cos command associates a class of service (CoS) to a list of traffic classes. Multiple commands create a complete traffic class to CoS map. The switch uses this map in CoS rewrite operations to fill the CoS field in outbound packets. This map is applicable to DSCP trusted ports and untrusted ports. CoS rewrite is disabled on CoS trusted ports. The show qos maps command displays the CoS to traffic class map.

The no qos traffic-class to cos and default qos traffic-class to cos commands restore the specified traffic class values to their default CoS settings by removing the corresponding qos map traffic-class to cos command from running-config.

Command Mode

Global Configuration

Command Syntax

qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to cos cos_value

no qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to cos

default qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to cos

Parameters
  • tc_num_x      Traffic class value. Value range varies by switch platform.
  • cos_value      Class of service (CoS) value. Value ranges from 0 to 7.

Default Inbound Traffic Class to CoS Map

Table 37 displays the default traffic class to CoS map for each platform.

Table 37. Default Traffic Class to CoS Rewrite Value Map
Traffic Class 0 1 2 3 4 5 6 7
CoS Rewrite Value (Arad and /Jericho) 1 0 2 3 4 5 6 7
CoS Rewrite Value (FM6000) 1 0 2 3 4 5 6 7
CoS Rewrite Value (Helix) 1 0 2 3 4 5 6 7
CoS Rewrite Value (Petra) 1 0 2 3 4 5 6 7
CoS Rewrite Value (Trident and Tomahawk) 1 0 2 3 4 5 6 7
CoS Rewrite Value (Trident II) 1 0 2 3 4 5 6 7
Example
This command assigns the CoS 2 to traffic classes 1, 3, and 5.
switch(config)#qos map traffic-class 1 3 5 to cos 2
switch(config)#

qos map traffic-class to dscp

The qos map traffic-class to dscp command associates a Differentiated Services Code Point (DSCP) value to a list of traffic classes. Multiple commands create a complete traffic class to DSCP map. The switch uses this map in DSCP rewrite operations to fill the DSCP field in outbound packets. This map is applicable to CoS trusted ports and untrusted ports but disabled by default on these ports. DSCP rewrite is disabled on DSCP trusted ports. The show qos maps command displays the traffic class to DSCP map.

The no qos traffic-class to dscp and default qos traffic-class to dscp commands restore the specified traffic class values to their default DSCP settings by removing the corresponding qos map traffic-class to dscp command from running-config.

Command Mode

Global Configuration

Command Syntax

qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to dscp dscp_value

no qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to dscp

default qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to dscp

Parameters
  • tc_num_x      Traffic class value. Value range varies by switch platform.
  • dscp_value      Differentiated services code point (DSCP) value. Value ranges from 0 to 63.

Default Inbound Traffic Class to DSCP Map

Table 38 displays the default traffic class to DSCP map for each platform.

Table 38. Default Traffic Class to DSCP Rewrite Value Map
Traffic Class 0 1 2 3 4 5 6 7
DSCP Rewrite Value (FM6000) 8 0 16 24 32 40 48 56
DSCP Rewrite Value (Helix) 8 0 16 24 32 40 48 56
DSCP Rewrite Value (Trident and Tomahawk) 8 0 16 24 32 40 48 56
DSCP Rewrite Value (Trident II) 8 0 16 24 32 40 48 56
Example
This command assigns the DSCP value of 17 to traffic classes 1, 2, and 4.
switch(config)#qos map traffic-class 1 2 4 to dscp 17
switch(config)#

qos map traffic-class to mc-tx-queue

The qos map traffic-class to mc-tx-queue command associates a multicast transmit queue to a list of traffic classes. Multiple commands create a complete traffic class to mc-tx-queue map. The switch uses this map to route outbound packets to transmit queues, which in turn schedules their transmission from the switch. The show qos maps command displays the traffic class to multicast transmit queue map.

The no qos traffic-class to mc-tx-queue and default qos traffic-class to mc-tx-queue commands restore the default traffic class to multicast transmit queue map for the specified traffic class values by removing the corresponding qos map traffic-class to mc-tx-queue command from running-config.

Command Mode

Global Configuration

Command Syntax

qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to mc-tx-queue mtq_value

no qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to mc-tx-queue

default qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to mc-tx-queue

Parameters
  • tc_num_x     Traffic class value. Value ranges from 0 to 7.
  • mtq_value     Multicast transmit queue number. Value ranges from 0 to 3.

Default Inbound Traffic Class to Multicast Transmit Queue Map

Table 39displays the default traffic class to multicast transmit queue map for Trident and Tomahawk platform switches.

Table 39. Default Traffic Class to Multicast Transmit Queue Map
Traffic Class 0 1 2 3 4 5 6 7
Multicast Transmit Queue (Trident and Tomahawk) 0 0 1 1 2 2 3 3
Related Commands
Example
This command maps traffic classes 0, 4, and 5 to mc-tx-queue 2.
switch(config)#qos map traffic-class 0 4 5 to mc-tx-queue 2
switch(config)#

qos map traffic-class to tx-queue

The qos map traffic-class to tx-queue command associates a transmit queue (tx-queue) to a list of traffic classes. Multiple commands create a complete traffic to tx-queue map. The switch uses this map to route outbound packets to transmit queues, which in turn schedules their transmission from the switch. The show qos maps command displays the transmit queue to traffic class map.

The no qos traffic-class to tx-queue and default qos traffic-class to tx-queue commands restore the specified traffic class values to their default transmit queue settings by removing the corresponding qos map traffic-class to tx-queue command from running-config.

Command Mode

Global Configuration

Command Syntax

qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to tx-queue txq_value

no qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to tx-queue

default qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to tx-queue

Parameters
  • tc_num_x      Traffic class value. Value range varies by platform.
  • txq_value      Transmit queue value. Value range varies by platform.

Restrictions

FM6000: When priority flow control (PFC) is enabled, traffic classes are mapped to their corresponding transmit queues, regardless of existing qos map traffic-class to tx-queue statements.

Arad, Jericho, and Petra: Traffic class 7 always maps to transmit queue 7. This association is not editable.

Default Inbound Traffic Class to Transmit Queue Map

Table 40 displays the transmit queue to traffic class map.

Table 40. Default Traffic Class to Transmit Queue Map
Traffic Class

 

0

1

2

3

4

5

6

7

Transmit Queue (Arad /Jericho) 0 1 2 3 4 5 6 7
Transmit Queue (FM6000) 0 1 2 3 4 5 6 7
Transmit Queue (Helix) 0 1 2 3 4 5 6 7
Transmit Queue (Petra) 0 1 2 3 4 5 6 7
Transmit Queue (Trident II) 0 1 2 3 4 5 6 7
Related Commands
Example
This command maps traffic classes 0, 4, and 5 to tx-queue 4.
switch(config)#qos map traffic-class 0 4 5 to tx-queue 4
switch(config)#

qos map traffic-class to uc-tx-queue

The qos map traffic-class to uc-tx-queue command associates a unicast transmit queue to a list of traffic classes. Multiple commands create a complete traffic class to unicast transmit queue map. The switch uses this map to route outbound packets to transmit queues, which in turn schedules their transmission from the switch. The show qos maps command displays the traffic class to unicast transmit queue map.

The no qos traffic-class to uc-tx-queue and default qos traffic-class to uc-tx-queue commands restore the default traffic class to unicast transmit queue map for the specified traffic class values by removing the corresponding qos map traffic-class to uc-tx-queue command from running-config.

Command Mode

Global Configuration

Command Syntax

qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to uc-tx-queue utq_value

no qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to uc-tx-queue

default qos map traffic-class tc_num_1 [tc_num_2 ... tc_num_n] to uc-tx-queue

Parameters
  • tc_num_x     Traffic class value. Value ranges from 0 to 7.
  • utq_value     Unicast transmit queue number. Value ranges from 0 to 7.

Default Inbound Traffic Class to Unicast Transmit Queue Map

Table 41 displays the default traffic class to Unicast transmit queue map for Trident and Tomahawk platform switches.

Table 41. Default Traffic Class to Unicast Transmit Queue Map
Traffic Class 0 1 2 3 4 5 6 7
Unicast Transmit Queue (Trident and Tomahawk) 0 1 2 3 4 5 6 7
Related Commands
Example
This command maps traffic classes 0, 4, and 5 to unicast transmit queue 4.
switch(config)#qos map traffic-class 0 4 5 to uc-tx-queue 4
switch(config)#

qos profile

The qos profile command places the switch in QoS profile configuration mode and for the specified profile and creates the profile if it does not already exist. QoS profiles are used to apply the same QoS configuration to multiple interfaces.

The no qos profile and default qos profile command deletes the QoS profile from the running configuration.

The exit command returns the switch to global configuration mode.

Command Mode

Global Configuration

Command Syntax

qos profile profile_name

no qos profile profile_name

default qos profile profile_name

Parameter

profile_name     QoS profile name.

Note: Commands use a subset of the listed fields. Available subset depends on the specified parameter. Use CLI syntax assistance to view options for specific parameter when creating a QoS profile.
Example

This command places the switch in QoS profile configuration mode for policy map policy map “TP” and creates the policy map if it does not already exist.

switch(config)#qos profile TP
switch(config-qos-profile-TP)#

qos random-detect ecn global-buffer (Helix)

The qos random-detect ecn global-buffer command enables ECN marking for globally shared packet memory and specifies minimum and maximum queue threshold sizes. Hosts can advertise their ECN capabilities in the ToS DiffServ field’s two least significant bits:
  • 00     Non ECN Capable transport.
  • 10     ECN Capable transport.
  • 01     ECN Capable transport.
  • 11     Congestion encountered.
Congestion is determined by comparing average queue size with queue thresholds. Average queue size is calculated through a formula based on the previous average and current queue size. Packets are marked based on this average size and the specified thresholds:
  • Average queue size below minimum threshold: Packets are queued normally.
  • Average queue size above maximum threshold: Packets are marked congestion encountered.
  • Average queue size between minimum and maximum thresholds. Packets are queued or marked congestion encountered. The proportion of marked packets varies linearly with average queue size:
    • 0% are marked when average queue size is less than or equal to minimum threshold.
    • 100% are marked when average queue size is greater than or equal to maximum threshold.

    When transmitted packets are marked Non ECN Capable, congestion packets are dropped, not marked.

The no qos random-detect ecn global-buffer and default qos random-detect ecn global-buffer commands disables ECN marking for the shared buffer by removing the qos random-detect ecn global-buffer command from running-config.

Command Mode

Global Configuration

Command Syntax

qos random-detect ecn global-buffer minimum-threshold MIN maximum-threshold MAX

no qos random-detect ecn global-buffer

default qos random-detect ecn global-buffer

Guidelines

Packet memory is divided into 46080 208-byte cells, whose allocation is managed by the memory management unit (MMU). The MMU tracks the cells that each entity uses and determines the number of cells that can be allocated to an entity.

Parameters

MIN and MAX parameters must use the same data unit.
  • MINMinimum threshold. Options include:
    • <1 to 19456> segments     208-byte segments units.
    • <1 to 4> mbytes     Megabyte units.
    • <1 to 4046> kbytes     Kilobyte units.
    • <1 to 4046848> bytes    Byte units.
  • MAXMaximum threshold. Options include:
    • <1 to 19456> segments     208-byte segments units.
    • <1 to 4> mbytes     Megabyte units.
    • <1 to 4046> kbytes     Kilobyte units.
    • <1 to 4046848> bytes    Byte units.

Related Command

random-detect ecn (Helix) enables ECN marking for a unicast transmit queue.

Examples
  • This command enables ECN marking of unicast packets from the global data pool and sets the minimum and maximum thresholds at 20 and 500 segments.
    switch(config)#qos random-detect ecn global-buffer minimum-threshold 20 segments 
    maximum-threshold 500 segments
    switch(config)#
  • This command disables ECN marking of unicast packets from the global data pool.
    switch(config)#no qos random-detect ecn global-buffer
    switch(config)#

qos random-detect ecn global-buffer (Trident and Tomahawk)

The qos random-detect ecn global-buffer command enables ECN marking for globally shared packet memory and specifies minimum and maximum queue threshold sizes. Hosts can advertise their ECN capabilities in the ToS DiffServ field’s two least significant bits:
  • 00     Non ECN Capable transport.
  • 10     ECN Capable transport.
  • 01     ECN Capable transport.
  • 11     Congestion encountered.
Congestion is determined by comparing average queue size with queue thresholds. Average queue size is calculated through a formula based on the previous average and current queue size. Packets are marked based on this average size and the specified thresholds:
  • Average queue size below minimum threshold: Packets are queued normally.
  • Average queue size above maximum threshold: Packets are marked congestion encountered.
  • Average queue size between minimum and maximum thresholds. Packets are queued or marked congestion encountered. The proportion of marked packets varies linearly with average queue size:
    • 0% are marked when average queue size is less than or equal to minimum threshold.
    • 100% are marked when average queue size is greater than or equal to maximum threshold.

    When transmitted packets are marked Non ECN Capable, congestion packets are dropped, not marked.

The no qos random-detect ecn global-buffer and default qos random-detect ecn global-buffer commands disables ECN marking for the shared buffer by removing the qos random-detect ecn global-buffer command from running-config.

Command Mode

Global Configuration

Command Syntax

qos random-detect ecn global-buffer minimum-threshold MIN maximum-threshold MAX

no qos random-detect ecn global-buffer

default qos random-detect ecn global-buffer

Guidelines

Packet memory is divided into 46080 208-byte cells, whose allocation is managed by the memory management unit (MMU). The MMU tracks the cells that each entity uses and determines the number of cells that can be allocated to an entity.

Parameters

MIN and MAX parameters must use the same data unit.
  • MINMinimum threshold. Options include:
    • <1 to 46080> segments     208-byte segments units.
    • <1 to 9> mbytes     Megabyte units.
    • <1 to 9584> kbytes     Kilobyte units.
    • <1 to 9584640> bytes    Byte units.
  • MAXMaximum threshold. Options include:
    • <1 to 46080> segments     208-byte segments units.
    • <1 to 9> mbytes     Megabyte units.
    • <1 to 9584> kbytes     Kilobyte units.
    • <1 to 9584640> bytes    Byte units.

Related Command

random-detect ecn (Trident and Tomahawk) enables ECN marking for a unicast transmit queue.

Examples
  • This command enables ECN marking of unicast packets from the global data pool and sets the minimum and maximum thresholds at 20 and 500 segments.
    switch(config)#qos random-detect ecn global-buffer minimum-threshold 20 segments maximum-threshold 500 segments
    switch(config)#
  • This command disables ECN marking of unicast packets from the global data pool.
    switch(config)#no qos random-detect ecn global-buffer
    switch(config)#

qos rewrite cos

The qos rewrite cos command enables the rewriting of the CoS field for outbound tagged packets that were received on DSCP trusted ports and untrusted ports. CoS rewrite is always disabled on CoS trusted ports. The CoS value that is written into the packet is based on the data stream’s traffic class. CoS rewriting is active by default.

The no qos rewrite cos command disables CoS rewriting on the switch. The default qos rewrite cos command restores the default setting of enabling CoS rewriting by removing the no qos rewrite cos command from running-config.

Command Mode

Global Configuration

Command Syntax

qos rewrite cos

no qos rewrite cos

default qos rewrite cos

Related Command

qos map traffic-class to cos configures the traffic class to CoS rewrite map.

Example

This command enables CoS rewrite.

switch(config)#qos rewrite cos
switch(config)#

qos rewrite dscp

The qos rewrite dscp command enables the rewriting of the DSCP field for outbound tagged packets that were received on CoS trusted ports and untrusted ports. DSCP rewrite is always disabled on DSCP trusted ports. The DSCP value that is written into the packet is based on the data stream’s traffic class. DSCP rewriting is disabled by default.

The no qos rewrite dscp and default qos rewrite dscp commands disable DSCP rewriting on the switch by removing the no qos rewrite dscp command from running-config.

Command Mode

Global Configuration

Command Syntax

qos rewrite dscp

no qos rewrite dscp

default qos rewrite dscp

Related Command

qos map traffic-class to dscp configures the traffic class to DSCP rewrite map.

Example
This command enables DSCP rewrite.
switch(config)#qos rewrite dscp
switch(config)#

qos trust

The qos trust command configures the quality of service port trust mode for the configuration mode interface. Trust-enabled ports classify traffic by examining the traffic’s CoS or DSCP value. Port trust mode default setting is cos for switched interfaces and dscp for routed interfaces.

The default qos trust command restores the default trust mode on the configuration mode interface by removing the corresponding qos trust or no qos trust statement from running-config.

The no qos trust command performs the following:
  • no qos trust places the port in untrusted mode.
  • no qos trust cos removes the corresponding qos trust cos statement.
  • no qos trust dscp removes the corresponding qos trust dscp statement.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

qos trust [MODE]

no qos trust [MODE]

default qos trust

Parameters
  • MODETrust mode assigned to the port. Options include:
    • cos      Enables cos trust mode.
    • dscp      Enables dscp trust mode.

no qos trust Enables untrusted mode on the port.

Examples
  • This command configures trust mode of dscp for Ethernet interface 7.
    switch(config)#interface Ethernet 7
    switch(config-if-Et7)#qos trust dscp
    switch(config-if-Et7)#show active
    interface Ethernet7
     qos trust dscp
    switch(config-if-Et7)#
  • This command configures trust mode of untrusted for Port Channel interface 23.
    switch(config)#interface port-channel 23
    switch(config-if-Po23)#no qos trust
    switch(config-if-Po23)#show active
    interface Port-Channel23
     no qos trust
    switch(config-if-Po23)#

random-detect ecn (Arad/Jericho)

The random-detect ecn command enables ECN marking for the configuration mode unicast transmit queue and specifies threshold queue sizes. Hosts can advertise their ECN capabilities in the ToS DiffServ field’s two least significant bits:
  • 00     Non ECN Capable transport.
  • 10     ECN Capable transport.
  • 01     ECN Capable transport.
  • 11     Congestion encountered.
Congestion is determined by comparing average queue size with queue thresholds. Average queue size is calculated through a formula based on the previous average and current queue size. Packets are marked based on this average size and the specified thresholds:
  • Average queue size below minimum threshold: Packets are queued normally.
  • Average queue size above maximum threshold: Packets are marked congestion encountered.
  • Average queue size between minimum and maximum thresholds. Packets are queued or marked congestion encountered. The proportion of marked packets varies linearly with average queue size:
    • 0% are marked when average queue size is less than or equal to minimum threshold.
    • 100% are marked when average queue size is greater than or equal to maximum threshold.

    When transmitted packets are marked Non ECN Capable, congestion packets are dropped, not marked.

The no random-detect ecn and default qos random-detect ecn commands disables ECN marking for the shared buffer by removing the qos random-detect ecn command from running-config.

Command Mode

Tx-Queue configuration

Command Syntax

random-detect ecn minimum-threshold MIN maximum-threshold MAX

no random-detect ecn

default random-detect ecn

Parameters

MIN and MAX parameters must use the same data unit.
  • MINMinimum threshold. Options include:
    • <1 to 256> mbytes     Megabyte units.
    • <1 to 256000> kbytes     Kilobyte units.
    • <1 to 256000000> bytes    Byte units.
  • MAXMaximum threshold. Options include:
    • <1 to 256> mbytes     Megabyte units.
    • <1 to 256000> kbytes     Kilobyte units.
    • <1 to 256000000> bytes    Byte units.

Related Command

tx-queue (Arad/Jericho) places the switch in tx-queue configuration mode.

Example
These commands enable ECN marking of unicast packets from unicast transmit queue 4 of Ethernet interface 3/5/1, setting thresholds at 128 kbytes and 1280 kbytes.
switch(config)#interface ethernet 3/5/1
switch(config-if-Et3/5/1)#tx-queue 4
switch(config-if-Et3/5/1-txq-4)#random-detect ecn minimum-threshold 128 kbytes 
maximum-threshold 1280 kbyte
switch(config-if-Et3/5/1-txq-4)#show active
interface Ethernet3/5/1
 tx-queue 4
random-detect ecn minimum-threshold 128 kbytes maximum-threshold 1280 kbytes
switch(config-if-Et3/5/1-txq-4)#

random-detect ecn (Helix)

The random-detect ecn command enables ECN marking for the configuration mode unicast transmit queue and specifies threshold queue sizes. Hosts can advertise their ECN capabilities in the ToS DiffServ field’s two least significant bits:
  • 00     Non ECN Capable transport.
  • 10     ECN Capable transport.
  • 01     ECN Capable transport.
  • 11     Congestion encountered.
Congestion is determined by comparing average queue size with queue thresholds. Average queue size is calculated through a formula based on the previous average and current queue size. Packets are marked based on this average size and the specified thresholds:
  • Average queue size below minimum threshold: Packets are queued normally.
  • Average queue size above maximum threshold: Packets are marked congestion encountered.
  • Average queue size between minimum and maximum thresholds. Packets are queued or marked congestion encountered. The proportion of marked packets varies linearly with average queue size:
    • 0% are marked when average queue size is less than or equal to minimum threshold.
    • 100% are marked when average queue size is greater than or equal to maximum threshold.

    When transmitted packets are marked Non ECN Capable, congestion packets are dropped, not marked.

    Average queue length is tracked for transmit queues and the global pool independently. When either entity reaches its maximum threshold, all subsequent packets are marked.

The no random-detect ecn and default random-detect ecn commands disable ECN marking on the configuration mode queue, deleting the corresponding random-detect ecn command from running-config.

Command Mode

Tx-Queue configuration

Command Syntax

random-detect ecn minimum-threshold MIN maximum-threshold MAX

no random-detect ecn

default random-detect ecn

Parameters

MIN and MAX parameters must use the same data unit.
  • MINMinimum threshold. Options include:
    • <1 to 46080> segments     208-byte segments units.
    • <1 to 9> mbytes      Megabyte units.
    • <1 to 9584> kbytes      Kilobyte units.
    • <1 to 9584640> bytes    Byte units.
  • MAX Maximum threshold. Options include:
    • <1 to 46080> segments     208-byte segments units.
    • <1 to 9> mbytes     Megabyte units.
    • <1 to 9584> kbytes     Kilobyte units.
    • <1 to 9584640> bytes    Byte units.
Related Commands
Examples
  • These commands enable ECN marking of unicast packets from transmit queue 4 of Ethernet interface 15, setting thresholds at 10 and 100 segments.
    switch(config)#interface ethernet 15
    switch(config-if-Et15)#uc-tx-queue 4
    switch(config-if-Et15-txq-4)#random-detect ecn minimum-threshold 10 segments 
    maximum-threshold 100 segments
    switch(config-if-Et15-txq-4)#show active
    interface Ethernet15
     tx-queue 4
    random-detect ecn minimum-threshold 10 segments maximum-threshold 100 
    segments
    switch(config-if-Et15-txq-4)#exit
    switch(config-if-Et15)
  • This command disables ECN marking of unicast packets from transmit queue 4 of Ethernet interface 15.
    switch(config-if-Et15-txq-4)#no random-detect ecn
    switch(config-if-Et15-txq-4)#show active
    interface Ethernet15
    switch(config-if-Et15-txq-4)#exit
    switch(config-if-Et15)#

random-detect ecn (Trident and Tomahawk)

The random-detect ecn command enables ECN marking for the configuration mode unicast transmit queue and specifies threshold queue sizes. Hosts can advertise their ECN capabilities in the ToS DiffServ field’s two least significant bits:
  • 00 Non ECN Capable transport.
  • 10     ECN Capable transport.
  • 01     ECN Capable transport.
  • 11     Congestion encountered.
Congestion is determined by comparing average queue size with queue thresholds. Average queue size is calculated through a formula based on the previous average and current queue size. Packets are marked based on this average size and the specified thresholds:
  • Average queue size below minimum threshold: Packets are queued normally.
  • Average queue size above maximum threshold: Packets are marked congestion encountered.
  • Average queue size between minimum and maximum thresholds. Packets are queued or marked congestion encountered. The proportion of marked packets varies linearly with average queue size:
    • 0% are marked when average queue size is less than or equal to minimum threshold.
    • 100% are marked when average queue size is greater than or equal to maximum threshold.

    When transmitted packets are marked Non ECN Capable, congestion packets are dropped, not marked.

    Average queue length is tracked for transmit queues and the global pool independently. When either entity reaches its maximum threshold, all subsequent packets are marked.

The no random-detect ecn and default random-detect ecn commands disable ECN marking on the configuration mode queue, deleting the corresponding random-detect ecn command from running-config.

Command Mode

Uc-Tx-Queue configuration

Command Syntax

random-detect ecn minimum-threshold MIN maximum-threshold MAX

no random-detect ecn

default random-detect ecn

Parameters

MIN and MAX parameters must use the same data unit.
  • MINMinimum threshold. Options include:
    • <1 to 46080> segments      208-byte segments units.
    • <1 to 9> mbytes      Megabyte units.
    • <1 to 9584> kbytes     Kilobyte units.
    • <1 to 9584640> bytes    Byte units.
  • MAX Maximum threshold. Options include:
    • <1 to 46080> segments     208-byte segments units.
    • <1 to 9> mbytes     Megabyte units.
    • <1 to 9584> kbytes      Kilobyte units.
    • <1 to 9584640> bytes     Byte units.
Related Commands
Examples
  • These commands enable ECN marking of unicast packets from unicast transmit queue 4 of Ethernet interface 15, setting thresholds at 10 and 100 segments.
    switch(config)#interface ethernet 15
    switch(config-if-Et15)#uc-tx-queue 4
    switch(config-if-Et15-uc-txq-4)#random-detect ecn minimum-threshold 10 segments 
    maximum-threshold 100 segments
    switch(config-if-Et15-uc-txq-4)#show active
    interface Ethernet15
     uc-tx-queue 4
    random-detect ecn minimum-threshold 10 segments maximum-threshold 100 
    segments
    switch(config-if-Et15-uc-txq-4)#exit
    switch(config-if-Et15)#
  • This command disables ECN marking of unicast packets from unicast transmit queue 4 of Ethernet interface 15.
    switch(config-if-Et15-uc-txq-4)#no random-detect ecn
    switch(config-if-Et15-uc-txq-4)#show active
    interface Ethernet15
    switch(config-if-Et15-uc-txq-4)#exit
    switch(config-if-Et15)#

service-policy type qos input

The service-policy type qos input command applies the specified policy map to a QoS profile. The profile is then applied to an interface in interface configuration mode using the service-profile command.

The no service-policy type qos and default service-policy type qos command deletes the policy map from the profile.

The exit command returns the switch to global configuration mode.

Command Mode

QoS Profile Configuration

Command Syntax

service-policy type qos input policy_map_name

no service-policy type qos input policy_map_name

default service-policy type qos input policy_map_name

Parameter
  • policy_map_name     QoS policy map name.
Example
This command applies the policy map PM-1 to the QoS profile TP.
switch(config-qos-profile-TP)#service-policy type qos input PM-1
switch(config-qos-profile-TP)#

service-profile

The command applies the QoS profile to the configuration mode interface.

The no service-profile and the default service-profile command removes the QoS profile from the interface.

The exit command returns the switch to global configuration mode.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

service-profile profile_name

no service-profile profile_name

default service-profile profile_name

Parameter

profile_name QoS profile name.

Example
This commands applies the QoS profile TP to Ethernet interface 13.
switch(config)#interface ethernet 13
switch(config-if-Et13)#service-profile TP

shape rate (Interface – Arad/Jericho)

The shape rate command specifies the maximum bandwidth for outbound traffic on the configuration mode interface, also known as queue shaping. The shape rate for individual transmit queues is configured by theshape rate (Tx-queue – Arad/Jericho) command. By default, outbound transmission rate is not bounded by a shape rate.

The no shape rate and default shape rate commands remove the shape rate bandwidth limit on the configuration mode interface by deleting the corresponding shape rate command from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

shape rate byte_limit [kbps]

no shape rate

default shape rate

Parameters

byte_limit      Shape rate applied to interface (Kbps). Value ranges from 162 to 100000000.

Example
This command configures a port shape rate of 5 Gbps on Ethernet interface 3/5/1.
switch(config)#interface ethernet 3/5/1
switch(config-if-Et3/5/1)#shape rate 5000000
switch(config-if-Et3/5/1)#show qos interfaces ethernet 3/5/1

Ethernet3/5/1:

 Port shaping rate: 5000012 / 5000000 kbps

TxBandwidth Shape RatePriorityECN
 Queue(percent)(units)
 -----------------------------------------------------
 7- / -- / -(-)SP / SPD
 6- / -- / -(-)SP / SPD
 5- / -- / -(-)SP / SPD
 4- / -- / -(-)SP / SPD
 3- / -- / -(-)SP / SPD
 2- / -- / -(-)SP / SPD
 1- / -- / -(-)SP / SPD
 0- / -- / -(-)SP / SPD


switch(config-if-Et3/5/1)#

shape rate (Interface – FM6000)

The shape rate command specifies the maximum bandwidth for outbound traffic on the configuration mode interface, also known as queue shaping. The shape rate for individual transmit queues is configured by the shape rate (Tx-queue – FM6000)command. By default, outbound transmission rate is not bounded by a shape rate.

The no shape rate and default shape rate commands remove the shape rate bandwidth limit on the configuration mode interface by deleting the corresponding shape rate command from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

shape rate byte_limit [kbps]

no shape rate

default shape rate

Parameters

byte_limit      Shape rate applied to interface (Kbps). Value ranges from 7000 to 10000000.

Guidelines

Enabling port shaping on an FM6000 interface disables queue shaping internally. Disabling port shaping restores queue shaping as specified in running-config.

Example
This command configures a port shape rate of 5 Gbps on Ethernet interface 5.
switch(config)#interface ethernet 5
switch(config-if-Et5)#shape rate 5000000
switch(config-if-Et5)#

shape rate (Interface – Helix)

The shape rate command specifies the maximum bandwidth for outbound traffic on the configuration mode interface, also known as queue shaping. The shape rate for individual transmit queues is configured by the shape rate (Interface – Helix)command. By default, outbound transmission rate is not bounded by a shape rate.

The no shape rate and default shape rate commands remove the shape rate bandwidth limit on the configuration mode interface by deleting the corresponding shape rate command from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

shape rate DATA_LIMIT

no shape rate

default shape rate

Parameters
  • DATA_LIMIT Shape rate applied to interface. Value range varies with data unit:
    • <8 to 40000000>      8 to 40000000 kbytes per second.
    • <8 to 40000000>kbps      8 to 40000000 kbytes per second.
    • <8 to 60000000>pps      8 to 60000000 packets per second.

Guidelines

Shaping rates of at least 8 kbps are supported. At shaping rates smaller than 1 Mbps, granularity and rounding errors may skew the actual shaping rate by 20% from the specified rate.

Example
This command configures a port shape rate of 5 Gbps on Ethernet interface 17.
switch(config)#interface ethernet 17
switch(config-if-Et17)#shape rate 5000000 kbps
switch(config-if-Et17)#show qos interface ethernet 17/3

Ethernet17:
 Trust Mode: COS
 Default COS: 0
 Default DSCP: 0

 Port shaping rate: 5000000 / 5000000 kbps

Tx Bandwidth Shape RatePriority
 Queue Guaranteed (units) (units)
 ------------------------------------------------------------
 7- / -(-) - / -(-)SP / SP
 6- / -(-) - / -(-)SP / SP

switch(config-if-Et17)#

shape rate (Interface – Petra)

The shape rate command specifies the maximum bandwidth for outbound traffic on the configuration mode interface, also known as queue shaping. The shape rate for individual transmit queues is configured by the shape rate (Tx-queue – Petra) command. By default, outbound transmission rate is not bounded by a shape rate.

The no shape rate and default shape rate commands remove the shape rate bandwidth limit on the configuration mode interface by deleting the corresponding shape rate command from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

shape rate data_limit [kbps]

no shape rate

default shape rate

Parameters
  • data_limit      Shape rate applied to interface (Kbps). Value ranges from 100 to 10000000.

Guidelines

The following port shaping rates are supported:
  • 1G ports: above 100 kbps.
  • 10G ports: above 7900 kbps.

Commands that specify a smaller shape rate disable port shaping on the interface.

Example
This command configures a port shape rate of 5 Gbps on Ethernet interface 3/3.
switch(config)#interface ethernet 3/3
switch(config-if-Et3/3)#shape rate 5000000
switch(config-if-Et3/3)#show active
interface Ethernet3/3
 shape rate 5000000
switch(config-if-Et3/3)#

shape rate (Interface – Trident and Tomahawk)

The shape rate command specifies the maximum bandwidth for outbound traffic on the configuration mode interface, also known as queue shaping. The shape rate for individual transmit queues is configured by the shape rate (Tx-queue – Trident and Tomahawk) command. By default, outbound transmission rate is not bounded by a shape rate.

The no shape rate and default shape rate commands remove the shape rate bandwidth limit on the configuration mode interface by deleting the corresponding shape rate command from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

shape rate DATA_LIMIT

no shape rate

default shape rate

Parameters
  • DATA_LIMITShape rate applied to interface. Value range varies with data unit:
    • <8 to 40000000>      8 to 40000000 kbytes per second.
    • <8 to 40000000>kbps      8 to 40000000 kbytes per second.
    • <8 to 60000000>pps      8 to 60000000 packets per second.

Guidelines

Shaping rates of at least 8 kbps are supported. At shaping rates smaller than 1 Mbps, granularity and rounding errors may skew the actual shaping rate by 20% from the specified rate.

Example
This command configures a port shape rate of 5 Gbps on Ethernet interface 5.
switch(config)#interface ethernet 5
switch(config-if-Et5)#shape rate 5000000
switch(config-if-Et5)#

shape rate (Interface – Trident II)

The shape rate command specifies the maximum bandwidth for outbound traffic on the configuration mode interface, also known as queue shaping. The shape rate for individual transmit queues is configured by the shape rate (Tx-queue – Trident II)command. By default, outbound transmission rate is not bounded by a shape rate.

The no shape rate and default shape rate commands remove the shape rate bandwidth limit on the configuration mode interface by deleting the corresponding shape rate command from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

shape rate DATA_LIMIT

no shape rate

default shape rate

Parameters
  • DATA_LIMITShape rate applied to interface. Value range varies with data unit:
    • <8 to 40000000>      8 to 40000000 kbytes per second.
    • <8 to 40000000>kbps      8 to 40000000 kbytes per second.
    • <8 to 60000000>pps      8 to 60000000 packets per second.

Guidelines

Shaping rates of at least 8 kbps are supported. At shaping rates smaller than 1 Mbps, granularity and rounding errors may skew the actual shaping rate by 20% from the specified rate.

Example
This command configures a port shape rate of 5 Gbps on Ethernet interface 17/3.
switch(config)#interface ethernet 17/3
switch(config-if-Et17/3)#shape rate 5000000 kbps
switch(config-if-Et17/3)#show qos interface ethernet 17/3

Ethernet17/3:
 Trust Mode: COS
 Default COS: 0
 Default DSCP: 0

 Port shaping rate: 5000000 / 5000000 kbps

Tx Bandwidth Shape RatePriority
 Queue Guaranteed (units) (units)
 ------------------------------------------------------------
 7- / -(-) - / -(-)SP / SP
 6- / -(-) - / -(-)SP / SP

switch(config-if-Et17/3)#

shape rate (Tx-queue – Arad/Jericho)

The shape rate command specifies the maximum bandwidth for outbound traffic on the transmit queue, also known as queue shaping. The shape rate for interfaces is configured by the shape rate (Interface – Arad/Jericho)command. By default, the configured outbound transmission rate is not bounded by a transmit queue shape rate.

Shaping rates greater than 50000 kbps are supported. At lower shaping rates (less than 10 Mbps), granularity and rounding errors may skew the actual shaping rate by 20% from the specified rate.

The no shape rate and default shape rate commands remove the shape rate bandwidth limit on the configuration mode queue by deleting the corresponding shape rate command from running-config.

Command Mode

Tx-Queue Configuration

Command Syntax

shape rate byte_limit [kbps]

no shape rate

default shape rate

Parameters

byte_limit     Shape rate applied to interface (Kbps). Value ranges from 50000 to 100000000.

Related Command

tx-queue (Arad/Jericho) places the switch in tx-queue configuration mode.

Example
These commands configure a shape rate of 1 Gbps on transmit queues 3 and 4 of Ethernet interface 3/4/1.
switch(config)#interface ethernet 3/4/1
switch(config-if-Et3/4/1)#tx-queue 4
switch(config-if-Et3/4/1-txq-4)#shape rate 1000000 kbps
switch(config-if-Et3/4/1-txq-4)#tx-queue 3
switch(config-if-Et3/4/1-txq-3)#shape rate 1000000 kbps
switch(config-if-Et3/4/1-txq-3)#show qos interface ethernet 3/4/1

Ethernet3/4/1:

 Port shaping rate: disabled

TxBandwidth Shape RatePriorityECN
 Queue(percent)(units)
 -----------------------------------------------------
 7- / -- / -(-)SP / SPD
 6- / -- / -(-)SP / SPD
 5- / -- / -(-)SP / SPD
 4- / -999 / 1000 ( Mbps ) SP / SPD
 3- / -999 / 1000 ( Mbps ) SP / SPD
 2- / -- / -(-)SP / SPD
 1- / -- / -(-)SP / SPD
 0- / -- / -(-)SP / SPD

switch(config-if-Et3/4/1-txq-3)#

shape rate (Tx-queue – FM6000)

The shape ratecommand specifies the maximum bandwidth for outbound traffic on the transmit queue, also known as queue shaping. The shape rate for interfaces is configured by the shape rate (Interface – FM6000) command. By default, the configured outbound transmission rate is not bounded by a transmit queue shape rate.

Queue shaping on an FM6000 port is supported only when port shaping is not enabled on the interface. Enabling port shaping on a port disables queue shaping internally. Disabling port shaping restores queue shaping as specified by running-config.

Shaping rates greater than 460 kbps are supported. At lower shaping rates (less than 10 Mbps), granularity and rounding errors may skew the actual shaping rate by 20% from the specified rate.

The no shape rate and default shape rate commands remove the shape rate bandwidth limit on the transmit queue by deleting the corresponding shape rate command from running-config.

Command Mode

Tx-Queue Configuration

Command Syntax

shape rate byte_limit [kbps]

no shape rate

default shape rate

Parameters

byte_limit      Shape rate applied to interface (Kbps). Value ranges from 464 to 10000000.

Related Commands
Example
These commands configure a shape rate of 1 Gbps (1,000,000 Kbps) on transmit queues 3 and 4 of Ethernet interface 19.
switch(config)#interface ethernet 19
switch(config-if-Et19)#tx-queue 4
switch(config-if-Et19-txq-4)#shape rate 1000000
switch(config-if-Et19-txq-4)#tx-queue 3
switch(config-if-Et19-txq-3)#shape rate 1000000
switch(config-if-Et19-txq-3)#show qos interface ethernet 19

Ethernet19:
 Trust Mode: COS

 Tx-Queue BandwidthShape Rate Priority
(percent) (Kbps)
 -----------------------------------------------
6 N/A disabledstrict
5 N/A disabledstrict
4 N/A1000000strict
3251000000 round-robin
225 disabled round-robin
125 disabled round-robin
025 disabled round-robin

switch(config-if-Et19-txq-3)#

shape rate (Tx-queue – Helix)

The shape rate command specifies the maximum bandwidth for outbound traffic on the transmit queue, also known as queue shaping. The shape rate for interfaces is configured by the shape rate (Interface – Helix) command. By default, the configured outbound transmission rate is not bounded by a transmit queue shape rate.

The no shape rate and default shape rate commands remove the shape rate bandwidth limit on the configuration mode transmit queue by deleting the corresponding shape rate command from running-config.

Command Mode

Tx-Queue Configuration

Command Syntax

shape rate byte_limit [kbps]

no shape rate

default shape rate

Parameters
  • DATA_LIMIT Shape rate applied to the queue. Value range varies with data unit:
    • <8 to 40000000>      8 to 40,000,000 kbytes per second.
    • <8 to 40000000>kbps      8 to 40000000 kbytes per second.
    • <8 to 60000000>pps      8 to 60000000 packets per second.

Restrictions

Queue shaping is not supported in cut-through mode.

Related Commands
Example
These commands configure a shape rate of 1 Gbps (1,000,000 Kbps) on transmit queues 3 and 4 of Ethernet interface 17/3.
switch(config)#interface ethernet 17/3
switch(config-if-Et17/3)#tx-queue 4
switch(config-if-Et17/3-txq-4)#shape rate 1000000 kbps
switch(config-if-Et17/3-txq-4)#tx-queue 3
switch(config-if-Et17/3-txq-3)#shape rate 1000000 kbps
switch(config-if-Et17/3-txq-3)#show qos interface ethernet 17/3

Ethernet17/3:

Tx Bandwidth Shape RatePriority
 Queue Guaranteed (units) (units)
 ------------------------------------------------------------
 7- / -(-) - / -(-)SP / SP
 6- / -(-) - / -(-)SP / SP
 5- / -(-) - / -(-)SP / SP
 4- / -(-) 1 / 1( Gbps ) SP / SP
 3- / -(-) 1 / 1( Gbps ) SP / SP
 2- / -(-) - / -(-)SP / SP
 1- / -(-) - / -(-)SP / SP
 0- / -(-) - / -(-)SP / SP

switch(config-if-Et17/3-txq-3)#

shape rate (Tx-queue – Petra)

The shape rate command specifies the maximum bandwidth for outbound traffic on the configuration mode transmit queue, also known as queue shaping. The shape rate for interfaces is configured by the shape rate (Interface – Petra) command. By default, the configured outbound transmission rate is not bounded by a transmit queue shape rate.

Queue shaping applies only to unicast traffic. Shaping rates of at least 162 Kbps are supported.

The no shape rate and default shape rate commands remove the shape rate bandwidth limit on the configuration mode queue by deleting the corresponding shape rate command from running-config.

Command Mode

Tx-Queue Configuration

Command Syntax

shape rate DATA_LIMIT

no shape rate

default shape rate

Parameters
  • DATA_LIMITShape rate applied to the queue. Value range varies with data unit:
    • <8 to 40000000>      8 to 40000000 kbytes per second.
    • <8 to 40000000>kbps      8 to 40000000 kbytes per second.
    • <8to 60000000>pps      8 to 60000000 packets per second.

    Shaping rates greater than 460 kbps are supported. At lower shaping rates (less than 10 Mbps), granularity and rounding errors may skew the actual shaping rate by 20% from the specified rate.

Related Commands
Example
These commands configure a shape rate of 1 Gbps (1,000,000 Kbps) on transmit queues 3 and 4 of Ethernet interface 3/28.
switch(config)#interface ethernet 3/28
switch(config-if-Et3/28)#tx-queue 4
switch(config-if-Et3/28-txq-4)#shape rate 1000000
switch(config-if-Et3/28-txq-4)#tx-queue 3
switch(config-if-Et3/28-txq-3)#shape rate 1000000
switch(config-if-Et3/28-txq-3)#show qos interface ethernet 3/28

Ethernet3/28:

 Tx-Queue BandwidthShape Rate Priority
(percent) (Kbps)
 -----------------------------------------------
7 N/A disabledstrict
6 N/A disabledstrict
5 N/A disabledstrict
4 N/A1000000strict
3251000000 round-robin
225 disabled round-robin
125 disabled round-robin
025 disabled round-robin

switch(config-if-Et3/28-txq-3)#

shape rate (Tx-queue – Trident and Tomahawk)

The shape rate command specifies the maximum bandwidth for outbound traffic on the configuration mode transmit queue, also known as queue shaping. The shape rate for interfaces is configured by the shape rate (Interface – Trident and Tomahawk) command. By default, the configured outbound transmission rate is not bounded by a transmit queue shape rate.

The no shape rate and default shape rate commands remove the shape rate limit from the configuration mode transmit queue by deleting the corresponding shape rate command from running-config.

Command Mode

Mc-Tx-Queue configuration

Uc-Tx-Queue configuration

Command Syntax

shape rate DATA_LIMIT

no shape rate

default shape rate

Parameters
  • DATA_LIMITShape rate applied to the queue. Value range varies with data unit:
  • <8 to 40000000>       8 to 40000000 kbytes per second.
  • <8 to 40000000>kbps       8 to 40000000 kbytes per second.
  • <8 to 60000000>pps      8 to 60000000 packets per second.

Related Commands

Guidelines

Shaping rates of at least 8 kbps are supported. At shaping rates smaller than 1 Mbps, granularity and rounding errors may skew the actual shaping rate by 20% from the specified rate.

When two queues source traffic from the same traffic class and the higher priority queue is shaped, that queue consumes all internal buffers, starving the lower priority queue even if bandwidth is available.

Example
These commands configure a shape rate of 1 Gbps (1,000,000 Kbps) on unicast transmit queues 3 and multicast transmit 4 of Ethernet interface 7.
switch(config)#interface ethernet 7
switch(config-if-Et7)#uc-tx-queue 3
switch(config-if-Et7-uc-txq-3)#shape rate 1000000
switch(config-if-Et7-uc-txq-3)#mc-tx-queue 2
switch(config-if-Et7-mc-txq-2)#shape rate 1000000
switch(config-if-Et7-mc-txq-2)#show qos interface ethernet 7

Ethernet7:

 Tx-Queue BandwidthShape Rate Priority Priority Group
(percent) (Kbps)
 ----------------------------------------------------------------
UC7 N/A disabledstrict1
UC6 N/A disabledstrict1
MC3 N/A disabledstrict1
UC5 N/A disabledstrict0
UC4 N/A disabledstrict0
MC2 N/A1000000strict0
UC3201000000 round-robin0
UC216 disabled round-robin0
MC116 disabled round-robin0
UC116 disabled round-robin0
UC016 disabled round-robin0
MC016 disabled round-robin0

switch(config-if-Et7-mc-txq-2)#

shape rate (Tx-queue – Trident II)

The shape rate command specifies the maximum bandwidth for outbound traffic on the configuration mode transmit queue, also known as queue shaping. The shape rate for interfaces is configured by the shape rate (Interface – Trident II)command. By default, the configured outbound transmission rate is not bounded by a transmit queue shape rate.

The no shape rate and default shape rate commands remove the shape rate bandwidth limit on the configuration mode transmit queue by deleting the corresponding shape rate command from running-config.

Command Mode

Tx-Queue Configuration

Command Syntax

shape rate byte_limit [kbps]

no shape rate

default shape rate

Parameters
  • DATA_LIMIT Shape rate applied to the queue. Value range varies with data unit:
    • <8 to 40000000>       8 to 40000000 kbytes per second.
    • <8 to 40000000>kbps      8 to 40000000 kbytes per second.
    • <8 to 60000000>pps      8 to 60000000 packets per second.

Restrictions

Queue shaping is not supported in cut-through mode

Related Commands
Example
These commands configure a shape rate of 1 Gbps (1,000,000 Kbps) on transmit queues 3 and 4 of Ethernet interface 17/3.
switch(config)#interface ethernet 17/3
switch(config-if-Et17/3)#tx-queue 4
switch(config-if-Et17/3-txq-4)#shape rate 1000000 kbps
switch(config-if-Et17/3-txq-4)#tx-queue 3
switch(config-if-Et17/3-txq-3)#shape rate 1000000 kbps
switch(config-if-Et17/3-txq-3)#show qos interface ethernet 17/3

Ethernet17/3:

Tx Bandwidth Shape RatePriority
 Queue Guaranteed (units) (units)
 ------------------------------------------------------------
 7- / -(-) - / -(-)SP / SP
 6- / -(-) - / -(-)SP / SP
 5- / -(-) - / -(-)SP / SP
 4- / -(-) 1 / 1( Gbps ) SP / SP
 3- / -(-) 1 / 1( Gbps ) SP / SP
 2- / -(-) - / -(-)SP / SP
 1- / -(-) - / -(-)SP / SP
 0- / -(-) - / -(-)SP / SP

switch(config-if-Et17/3-txq-3)#

show platform petraA traffic-class

The show platform petraA traffic-class command displays the traffic class assignment on all specified Petra chips. Each chip controls eight Ethernet interfaces. The default traffic class of an interface is specified by the traffic class assigned to the chip that controls the interface.

Traffic class assignments are configured with the platform petraA traffic-class command.

Valid command options include:
  • show platform petraA traffic-class       Traffic class of all chips on all linecard.
  • show platform petraA CHIP_NAME traffic-class      Traffic class of specified chip.
  • show platform petraA MODULE_NAME traffic-class   Traffic class of all chips on specified linecard.

Command Mode

EXEC

Command Syntax

show platform petraA traffic-class

show platform petraA CHIP_NAME traffic-class

show platform petraA MODULE_NAME traffic-class

Parameters
  • CHIP_NAMEName of Petra chip on linecard that control Ethernet ports. Options include:
    • petra cardX / chipY      All ports on PetraA chip chipY on linecard cardX (7500 Series).
    • petra chipZ     All ports on PetraA chip chipZ (7048 Series).

    7500 Series

    Switches can contain up to eight linecards. cardX varies from 3 to 10.

    Each linecard contains six PetraA chips. Each chip controls eight ports. chipY varies from 0 to 5:
    • 0 controls ports 1 through 8
    • 1 controls ports 9 through 16
    • 2 controls ports 17 through 24
    • 3 controls ports 25 through 32
    • 4 controls ports 33 through 40
    • 5 controls ports 41 through 48

    7048 Series

    Each switch contains two PetraA chips. chipZ varies from 0 to 1:
    • 0 controls ports 1 through 32
    • 1 controls ports 33 through 52
  • MODULE_NAME Name and number of linecard (7500 Series). Options include:
    • module linecard mod_num     Linecard number (3 to 10).
    • module mod_num      Linecard number (3 to 10).

Related Command

platform petraA traffic-class configures the default traffic class used by all ports on a specified chip.

Example
This command displays the traffic class of all chips on linecard 3.
switch#show platform petraA module linecard 3 traffic-class
Petra3/0 traffic-class: 1
Petra3/1 traffic-class: 1
Petra3/2 traffic-class: 1
Petra3/3 traffic-class: 1
Petra3/4 traffic-class: 5
Petra3/5 traffic-class: 1
switch#

show platform trident tcam qos detail

The show platform trident tcam qos detail command displays the list of all the SVIs that are sharing the TCAM entries.

Command Mode

EXEC

Command Syntax

show platform trident tcam qos detail

Example
This command displays the list of all the SVIs that are sharing the TCAM entries.
switch(config)#show platform trident tcam qos detail

=== Policy-map p01 type qos on switch Linecard0/0 ===
Interfaces : Vlan2 Vlan1
=-= Interface BitMap =-=
0x000000000000000001FFFFFE

show platform trident tcam shared vlan interface-class-id

The show platform trident tcam shared vlan interface-class-id command displays what SVIs are currently sharing the QoS policy-map in the below output under QoS PMAP Data.

Command Mode

EXEC

Command Syntax

show platform trident tcam shared vlan interface-class-id

Example
This command displays what SVIs are currently sharing the QoS policy-map in the below output under QoS PMAP Data.
switch(config)#show platform trident tcam shared vlan interface-class-id

=== Shared RACL Data on switch Linecard0/0 ===
=== Shared QoS Policy-map Data on switch Linecard0/0 ===
Interface Class Id VLANs
1 1 2

show platform xp qos tcam hit

The show platform xp qos tcam hit command displays the TCAM entries programmed for each policy-map as well as the traffic hits. The hits option is used to see the TCAM entries with nonzero traffic hits.

Command Mode

EXEC

Command Syntax

show platform xp qos tcam hit

Example
This command displays the QoS TCAM hits on Ethernet interface 10/1.
switch#show platform xp qos tcam hit

=== Policy-map test type qos on switch 0 ===
Assigned to ports: Ethernet10/1
=-= Class-map test type qos =-=
=== ACL test
=========================================================================================
|Seq|AclId|Prot|Port|SPort|Ecn|FFlg|DPort|Vlan|Action|Hits|Src Ip|Dest Ip|hwId| | | || | | |dscp|cos|tc|PolId| | | |
=========================================================================================
| 10| 0x01||| | |0x04| || 4|- | -| - |91852787| | | | 0 | 0x00 | | | | |0xfb|| || | |
-----------------------------------------------------------------------------------------

show policy-map interface

The show policy-map interface command displays contents of the policy map applied to specified the interface.

Command Mode

EXEC

Command Syntax

show policy-map interface interface_name

Parameters
  • interface_name    Interface for which command returns data. Options include:
    • no parameter      Returns data for all interfaces.
    • ethernet e_range      Ethernet interfaces specified by e_range.
    • port-channel p_range      Port channel interfaces specified by p_range.
Example
This command displays the name and contents of the policy map applied to Ethernet interface 1.
switch#show policy-map interface ethernet 1
Service-policy input: p1
 Hardware programming status: Successful

 Class-map: c2001 (match-any)
Match: vlan 2001 0xfff
 set dscp 4

 Class-map: c2002 (match-any)
Match: vlan 2002 0xfff
 set dscp 8

 Class-map: c2003 (match-any)
Match: vlan 2003 0xfff
 set dscp 12

show policy-map

The show policy-map command displays the policy map information for the configured policy map.

Command Mode

EXEC

Command Syntax

show policy-map policy_map_name [counters][interface | summary]

Parameters
  • policy_map_name QoS policy map name.
  • counters     Specifies the policy map traffic match count (This parameter is applicable only on DCS-7010, DCS-7050X, DCS7250X, DCS-7300X and DCS-7280(E/R), DCS-7500(E/R) series switches.)
  • interface      Specifies the service policy on an interface.
  • summary      Policy map summary.
Examples
  • The show policy-map command displays the information for the policy map policy1.
    switch#show policy-map policy1
    Service-policy policy1
    Class-map: class1 (match-any)
    Match: ip access-group name acl1
    Police cir 512000 bps bc 96000 bytes
    Class-map: class-default (match-any)
  • The show policy-map counters command displays the policy map traffic match count for the policy map configured.
    switch#show policy-map policy1 counters
    
    Service-policy input: policy1
     Hardware programming status: Successful
     Class-map: class1 (match-any)
    Match: vlan 20-40,1000-1250
     police rate 100 mbps burst-size 100 kbytes
    Interface: Ethernet16/1
    Conformed 28621 packets, 7098008 bytes -------------- packet match count
     
    Class-map: class-default (match-any)
    Matched Packets: 19 -------------- packet match count

show qos interfaces random-detect ecn

The show qos interfaces random-detect ecn command displays the Explicit Congestion Notification (ECN) configuration for each transmit queue on the specified interfaces.

Command Mode

EXEC

Command Syntax

show qos interfaces [INTERFACE_NAME] random-detect ecn

Parameters
  • INTERFACE_NAMEInterface for which command returns data. Options include:
    • <no parameter>     Returns data for all interfaces.
    • ethernet e_range      Ethernet interfaces specified by e_range.
    • port-channel p_range     Port-Channel Interfaces specified by p_range.
Example
This command configures ECN parameters for transmit queues 0 through 3 on Ethernet interface 3/5/1, then displays that configuration.
switch(config)#interface ethernet 3/5/1
switch(config-if-Et3/5/1)#tx-queue 0
switch(config-if-Et3/5/1-txq-0)#random-detect ecn minimum-threshold 2560 kbytes 
maximum-threshold 256000 kbytes
switch(config-if-Et3/5/1-txq-0)#tx-queue 1
switch(config-if-Et3/5/1-txq-1)#random-detect ecn minimum-threshold 25600 kbytes 
maximum-threshold 128000 kbytes
switch(config-if-Et3/5/1-txq-1)#tx-queue 2
switch(config-if-Et3/5/1-txq-2)#random-detect ecn minimum-threshold 25600 bytes 
maximum-threshold 128000 bytes
switch(config-if-Et3/5/1-txq-2)#tx-queue 3
switch(config-if-Et3/5/1-txq-3)#random-detect ecn minimum-threshold 25 mbytes 
maximum-threshold 128 mbytes
switch(config-if-Et3/5/1-txq-3)#show qos interfaces ethernet 3/5/1 random-detect 
ecn

Ethernet3/5/1:

 Tx-Queue Mininimum ThresholdMaximum ThresholdThreshold Unit
 ---------------------------------------------------------------------
7 -- -
6 -- -
5 -- -
4 -- -
325128 mbytes
2 25600 128000bytes
1 25600 128000 kbytes
02560 256000 kbytes

switch(config-if-Et3/5/1-txq-3)#

show qos interfaces trust

The show qos interfaces trust command displays the configured and operational QoS trust mode of all specified interfaces.

Command Mode

EXEC

Command Syntax

show qos interfaces [INTERFACE_NAME] trust

Parameters
  • INTERFACE_NAME Interface for which command returns data. Options include:
    • <no parameter>     Returns data for all interfaces.
    • ethernet e_range     Ethernet interfaces specified by e_range.
    • port-channel p_range     Port-Channel Interfaces specified by p_range.
Example
These commands configure a variety of QoS trust settings on a set of interfaces, then displays the QoS trust mode on these interfaces.
switch(config)#interface ethernet 1/1
switch(config-if-Et1/1)#qos trust cos
switch(config-if-Et1/1)#interface ethernet 1/2
switch(config-if-Et1/2)#qos trust dscp
switch(config-if-Et1/2)#interface ethernet 1/3
switch(config-if-Et1/3)#no qos trust
switch(config-if-Et1/3)#interface ethernet 1/4
switch(config-if-Et1/4)#default qos trust
switch(config-if-Et1/4)#interface ethernet 2/1
switch(config-if-Et2/1)#no switchport
switch(config-if-Et2/1)#default qos trust
switch(config-if-Et2/1)#show qos interface ethernet 1/1 - 2/4 trust

Port Trust Mode
 Operational Configured
---------------------------------------------------------------
Ethernet1/1COS COS
Ethernet1/2DSCPDSCP
Ethernet1/3UNTRUSTED UNTRUSTED
Ethernet1/4COS DEFAULT
Ethernet2/1DSCPDEFAULT
Ethernet2/2COS DEFAULT
Ethernet2/3COS DEFAULT
Ethernet2/4COS DEFAULT

switch(config-if-Et2/1)#

show qos interfaces

The show qos interfaces command displays the QoS, DSCP, and transmit queue configuration on a specified interface. Information provided by this command includes the ports trust setting, the default CoS value, and the DSCP value.

Command Mode

EXEC

Command Syntax

show qos interfaces INTERFACE_NAME

Parameters
  • INTERFACE_NAME Interface For which command returns data. Options include:
    • <no parameter>     Returns data for all interfaces.
    • ethernet e_num      Ethernet interface specified by e_num.
    • port-channel p_num Port-Channel Interface specified by p_num.
Example
This command lists the QoS configuration for Ethernet interface 4.
switch>show qos interfaces ethernet 4

Ethernet4:
 Trust Mode: COS
 Default COS: 0
 Default DSCP: 0

 Port shaping rate: 5000000Kbps

 Tx-Queue BandwidthShapeRatePriority
(percent) (Kbps)
 -----------------------------------------------
050 disabled round-robin
150 disabled round-robin
2 N/A disabledstrict
3 N/A1000000strict
4 N/A1000000strict
5 N/A1500000strict
6 N/A2000000strict

switch>

show qos maps

The show qos maps command lists the number of traffic classes that the switch supports and displays the CoS-Traffic Class, DSCP-Traffic Class, Traffic Class-CoS, and Traffic Class-Transmit Queue maps.

Command Mode

EXEC

Command Syntax

show qos maps Examples

Example
This command displays the QoS maps that are configured on the switch.
switch>show qos maps
Number of Traffic Classes supported: 8
   Number of Transmit Queues supported: 8
   Cos Rewrite:  Disabled
   Dscp Rewrite:  Disabled

   Cos-tc map:
     cos:  0  1  2  3  4  5  6  7
     ----------------------------
     tc:   1  0  2  3  4  5  6  7

 Dscp-tc map:
 d1 :d2 0123456789
 --------------------------------------
0 :     1111111100
1 : 0000002222
2 : 2222333333
3 : 3344444444
4 : 5555555566
5 : 6666667777
6 : 7777

 Tc-cos map:
 tc: 01234567
 ----------------------------
 cos:10234567

 Tc-dscp map:
 tc:01234567
 -----------------------------
 dscp:80 16 24 32 40 48 56

 Tc - tx-queue map:
 tc:01234567
 ---------------------------------
 tx-queue:01234567

switch>

show qos profile summary

The show qos profile summary command displays the QoS profile summary of those which are part of the running configuration.

Command Mode

EXEC

Command Syntax

show qos profile summary

Example
This command shows a summary of all QoS profiles configured on the switch.
switch(config)#show qos profile summary
Qos Profile: p
Configured on: Et13,7
Fabric
Po12
Qos Profile: p2
Configured on: Et56

show qos profile

The show qos profile command displays the contents of the specified QoS profile or of all QoS profiles in the running configuration.

Command Mode

EXEC

Command Syntax

show qos profile profile_name

Parameter
  • profile_nameQoS profile name.
Examples
  • This command displays the contents of all QoS profiles configured on the switch.
    switch(config)#show qos profile
    qos profile p
    qos cos 1
    no priority-flow-control pause watchdog
    priority-flow-control priority 1 no-drop
    priority-flow-control priority 2 no-drop
    qos profile p2
    qos cos 3
    priority-flow-control priority 0 no-drop
  • This command displays the configuration attached and information specific toQoS profile p2.
    switch#show qos profile p2
    qos profile p2
    qos cos 3
    priority-flow-control priority 0 no-drop

show qos random-detect ecn

show qos random-detetct ecn

Thecommand displays the global Explicit Congestion Notification (ECN) configuration.

Command Mode

EXEC

Command Syntax

show qos random-detect ecn

Example
These commands configure global ECN parameters, then displays that configuration.
switch(config)#qos random-detect ecn global-buffer minimum-threshold 2 mbytes 
maximum-threshold 5 mbytes
switch(config)#show qos random-detect ecn
 Minimum Threshold:2
 Maximum Threshold:5
 Threshold Unit:mbytes

switch(config)#

show run|grep sharing

Theshow run|grep sharing command displays whether the QoS policy-map sharing on SVIs is enabled or disabled.

Command Mode

EXEC

Command Syntax

show run|grep sharing

Example
This command displays whether the QoS policy-map sharing on SVIs is enabled or disabled.
switch#show run|grep sharing
hardware access-list qos resource sharing vlan in ---- 
If this message is displayed then QoS policy-map sharing on SVIs is enabled.

tx-queue (Arad/Jericho)

The tx-queue command places the switch in Tx-queue configuration mode to configure a transmit queue on the configuration mode interface. Tx-queue configuration mode is not a group change mode; running-config is changed immediately after commands are executed. The exit command does not affect the configuration.

Arad and Jericho platform switches have eight queues, 0 through 7, and all queues are exposed through the CLI. However, queue 7 is not user-configurable. Queue 7 is always mapped to traffic class 7, which is reserved for control traffic.

The exit command returns the switch to the configuration mode for the base Ethernet or port channel interface.

The no tx-queue and default tx-queue commands remove the configuration for the specified transmit queue by deleting all corresponding tx-queue mode statements from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

tx-queue queue_level

Parameters
  • queue_level The transmit queue. Values range from 0 to 7.

Guidelines

Arad and Jericho platform switch queues handle unicast traffic. Queues for multicast traffic are not supported.

Example
This command enters Tx-queue configuration mode for transmit queue 4 of Ethernet interface 3/3/3.
switch(config)#interface ethernet 3/3/3
switch(config-if-Et3/3/3)#tx-queue 4
switch(config-if-Et3/3/3-txq-4)#

tx-queue (FM6000)

The tx-queue command places the switch in Tx-queue configuration mode to configure a transmit queue on the configuration mode interface. Tx-queue configuration mode is not a group change mode; running-config is changed immediately after commands are executed. The exit command does not affect the configuration.

FM6000 platform switches have eight queues, 0 through 7. All queues are exposed through the CLI and are user configurable.

The exit command returns the switch to the configuration mode for the base Ethernet or port channel interface.

The no tx-queue and default tx-queue commands remove the configuration for the specified transmit queue by deleting the all corresponding tx-queue mode commands from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

tx-queue queue_level

Parameters

queue_level The transmit queue. Values range from 0 to 7.

Commands Available in tx-queue Configuration Mode

Guidelines

FM6000 platform switch queues handle unicast and multicast traffic.

Example
This command enters Tx-queue configuration mode for transmit queue 3 of Ethernet interface 5.
switch(config)#interface ethernet 5
switch(config-if-Et5)#tx-queue 3
switch(config-if-Et5-txq-3)#

tx-queue (Helix)

The tx-queue command places the switch in Tx-queue configuration mode to configure a transmit queue on the configuration mode interface. Tx-queue configuration mode is not a group change mode; running-config is changed immediately after commands are executed. The exit command does not affect the configuration.

Helix platform switches have eight unicast (UC0 – UC7) and eight multicast (MC0 – MC7) queues. Each UCx-MCx queue set is combined into a single queue group (L1.x), which is exposed to the CLI through this command.

The exit command returns the switch to the configuration mode for the base Ethernet or port channel interface.

The no tx-queue and default tx-queue commands remove the configuration for the specified transmit queue by deleting the all corresponding tx-queue mode commands from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax tx-queue queue_level

Parameters

queue_level Transmit queue group number. Values range from 0 to 7.

Commands Available in tx-queue Configuration Mode

Guidelines

Helix platform switch queues handle unicast and multicast traffic.

Example
This command enters Tx-queue configuration mode for transmit queue 4 of Ethernet interface 17/3.
switch(config)#interface ethernet 17/3
switch(config-if-Et17/3)#tx-queue 4
switch(config-if-Et17/3-txq-4)#

tx-queue (Petra)

The tx-queue command places the switch in Tx-queue configuration mode to configure a transmit queue on the configuration mode interface. Tx-queue configuration mode is not a group change mode; running-config is changed immediately after commands are executed. The exit command does not affect the configuration.

Petra platform switches have eight queues, 0 through 7, and all queues are exposed through the CLI. However, queue 7 is not user-configurable. Queue 7 is always mapped to traffic class 7, which is reserved for control traffic.

The exit command returns the switch to the configuration mode for the base Ethernet or port channel interface.

The no tx-queue and default tx-queue commands remove the configuration for the specified transmit queue by deleting the all corresponding tx-queue mode commands from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

tx-queue queue_level

Parameters

queue_level The transmit queue. Values range from 0 to 7.

Commands Available in tx-queue Configuration Mode

Guidelines

Petra platform switch queues handle unicast traffic. Queues for multicast traffic are not supported.

Example
This command enters Tx-queue configuration mode for transmit queue 3 of Ethernet interface 3/3.
switch(config)#interface ethernet 3/3
switch(config-if-Et3/3)#tx-queue 3
switch(config-if-Et3/3-txq-3)#

tx-queue (Trident II)

The tx-queue command places the switch in Tx-queue configuration mode to configure a transmit queue on the configuration mode interface. Tx-queue configuration mode is not a group change mode; running-config is changed immediately after commands are executed. The exit command does not affect the configuration.

Trident II platform switches have eight unicast (UC0 – UC7) and eight multicast (MC0 – MC7) queues. Each UCx-MCx queue set is combined into a single queue group (L1.x), which is exposed to the CLI through this command.

The exit command returns the switch to the configuration mode for the base Ethernet or port channel interface.

The no tx-queue and default tx-queue commands remove the configuration for the specified transmit queue by deleting the all corresponding tx-queue mode commands from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

tx-queue queue_level

Parameters

queue_level Transmit queue group number. Values range from 0 to 7.

Commands Available in tx-queue Configuration Mode

Trident II platform switch queues handle unicast and multicast traffic.

Example
This command enters Tx-queue configuration mode for transmit queue 4 of Ethernet interface 17/3.
switch(config)#interface ethernet 17/3
switch(config-if-Et17/3)#tx-queue 4
switch(config-if-Et17/3-txq-4)#

uc-tx-queue

The uc-tx-queue command places the switch in uc-tx-queue configuration mode to configure a unicast transmit queue on the configuration mode interface. Uc-tx-queue configuration mode is not a group change mode; running-config is changed immediately after commands are executed. The exit command does not affect the configuration.

Trident and Tomahawk switches have eight unicast queues (UC0 – UC7) and four multicast queues (MC0 – MC03), categorized into two priority groups. All queues are exposed through the CLI and are user-configurable.

  • Priority Group 1: UC7, UC6, MC3
  • Priority Group 0: UC5, UC4, MC2, UC3, UC2, MC1, UC1, UC0, MC0

The exitcommand returns the switch to the configuration mode for the base Ethernet or port channel interface.

The no uc-tx-queue and default uc-tx-queue commands remove the configuration for the specified transmit queue by deleting the all corresponding uc-tx-queue mode commands from running-config.

Command Mode

Interface-Ethernet Configuration

Interface-Port-Channel Configuration

Command Syntax

uc-tx-queue queue_level

Parameters

queue_level The multicast transmit queue number. Values range from 0 to 7.

Related Command

mc-tx-queue: Configures multicast transmit queues on Trident and Tomahawk platform switches.

Example
This command enters mc-tx-queue configuration mode for multicast transmit queue 4 of Ethernet interface 5.
switch(config)#interface ethernet 5
switch(config-if-Et5)#uc-tx-queue 4
switch(config-if-Et5-mc-txq-4)#
..