multicast-replication

Use the evpn irb option to enable inter-VLAN multicast forwarding in an EVPN-VXLAN network in one of the following modes (these options are mutually exclusive):

• local-only—Use this mode with an edge-routed bridging overlay, also known as a collapsed IP fabric. In this mode, the PFE on the leaf devices in the fabric performs local multicast routing at the fabric edge. The spine devices, also called lean spines, primarily act as IP transit devices for the fabric.

• local-remote—Use this mode with a centrally-routed bridging (CRB) overlay, also known as a two-layer IP fabric. In this mode, the spine devices in the fabric centrally route the multicast traffic between VLANS. The spine devices forward the routed VLAN traffic into the EVPN core toward interested receivers. The spine devices use a PIM designated router (DR) to avoid duplicating packets into the core. The leaf devices forward multicast traffic received on a VLAN to their receivers on that VLAN.

• oism—Use this mode with an ERB overlay to also support routing multicast traffic between the fabric and external devices. This mode implements optimized inter-subnet multicast (OISM) according to the IETF draft https://tools.ietf.org/html/draft-ietf-bess-evpn-irb-mcast. In this mode, the leaf devices operate in local-remote mode like in a CRB overlay design, while at the same time, the device performs local multicast routing. This mode enables the original regular OISM mode that uses a symmetric bridge domains (VLANs) OISM model. With regular OISM mode, you must configure all revenue bridge domains (VLANs) symmetrically on all OISM devices.

  OISM also supports:

  • Selective forwarding on the source VLAN.

  • Routing multicast traffic through an external PIM domain.

  Note:

  You can alternatively enable the enhanced version of OISM, if all of the OISM devices in your network support the enhanced OISM mode option. See the enhanced-oism option description (next) for information on when you might use that option instead of this original OISM mode option.

• enhanced-oism—This mode implements an enhanced version of OISM for EVPN multicast with ERB overlays. Enhanced OISM doesn't require you to configure all revenue bridge domains (VLANs) in the network on all OISM devices. On each device, you can configure only the revenue VLANs the device hosts. As a result, we say this mode uses an asymmetric bridge domains (VLANs) model. On supported devices, this mode enables OISM to scale well when your network has leaf devices that host larger numbers of different VLANs.

  This mode has some operational differences and small configuration differences to support the enhanced OISM asymmetric bridge domains model. As a result, if you enable OISM, all OISM devices in the network must use the same OISM mode option, either oism or enhanced-oism.

  You can choose to use the enhanced-oism option instead of the regular oism option if all OISM devices in the network support enhanced OISM. In that case, you might want to use enhanced OISM when:

  • Your network has a large number of revenue bridge domains (VLANs), and resources might be strained on some devices to configure all the VLANs there.

  • Your network has a large number of disjointed bridge domains (VLANs) in the network (different devices host different sets of VLANs).

  • On OISM devices in your network, you don't have policies configured that are based on the source MAC address of the packets. If you do have source MAC address policies, use the oism option in your network instead.

  Enhanced OISM also supports:

  • Selective forwarding on the source VLAN.

  • Routing multicast traffic through an external PIM domain.

Default EVPN multicast IRB mode: evpn irb local-remote

You can enable only one of these modes at a time, and you must configure the same mode on all devices in the network that process multicast traffic.

Configures a limit for the number of SMET next hops for selective multicast forwarding in an EVPN network. A PE device uses the SMET next hops list of outgoing interfaces to selectively replicate and forward multicast traffic. When the list of SMET next hops reaches this limit, the PE device stops adding new SMET next hops. At that point, the PE device sends new multicast group traffic to all egress devices.

• Range: 10,000 through 40,000

• Default: 10,000

Complete ingress replication of the multicast data packets where all the egress Packet Forwarding Engines receive packets from the ingress Packet Forwarding Engines directly.

Complete parallel replication of the multicast data packets.