In conventional VXLAN deployments, each MLAG pair of switches are represented as a common logical VTEP. VXLAN traffic can be decapsulated on either switch. In some networks, there are hosts that are singly connected to one of the MLAG pair. VXLAN packets destined for the singly connected host could land on the other MLAG peer and subsequently be forwarded over the MLAG peer-link to reach the destination host. This path is undesirable since it would use up some bandwidth on the peer-link.