NSX Data Center 6.4.2 and later supports multicast routing.
NSX uses two multicast routing protocols: Internet Group Management Protocol (IGMPv2) and Protocol Independent Multicast (PIM). PIM sparse mode is supported (PIM-SM). PIM is used on ESGs, but not on the DLR.
- Multicast routing is supported between the ESG and the DLR.
- Receiving hosts advertise their group membership to a local multicast router, enabling them to join and leave multicast groups.
- Protocol-Independent Multicast (PIM) is used for a router-to-router signaling. It maintains the current IP multicast service mode of receiver-initiated membership.
After any routing protocol is first enabled (or disabled and re-enabled), traffic is not forwarded until the protocol has converged, and the routes corresponding to the traffic have been learned and installed. In a multicast network, traffic forwarding requires both the unicast and the multicast routing protocols to converge. PIM Sparse-mode also requires that the Rendezvous Point (RP) for a multicast group is known before any control or data traffic for the multicast group is processed. When the PIM Bootstrap mechanism is used to disseminate the RP information, the Candidate RPs are learned only after a Bootstrap message from a PIM neighbor is received. These messages have an RFC default periodicity of 60 secs. If a Static RP is configured, the RP information is available immediately and the delay associated with the Bootstrap mechanism is avoided.
- IPv4 support.
- IGMPv2 support.
- PIM Sparse mode is supported.
- Rendezvous point information can be delivered through bootstrap messages, or statically configured.
- Replication Multicast Range should not overlap with a Transport Zone multicast range.
- An Edge Services Gateway (ESG) cannot be a Bootstrap Candidate Router.
- An ESG cannot be the Rendezvous Point (RP).
- The routes of multicast participating nodes must be learned explicitly either by the unicast routing protocol or through a static route. NSX does not use the default route for multicast reverse path forwarding (RPF) checks.
- During vMotion of virtual machines that are receivers of multicast, a 1–2 second multicast traffic loss can occur.
- Starting in NSX 6.4.7, distributed firewall (DFW) is supported for multicast traffic. However, IPFIX is not supported for multicast.
- Starting in NSX 6.4.7, edge firewall is supported for multicast traffic. Edge firewall supports filtering of IGMP packets on the basis of protocol in the IP header. The firewall cannot filter the type of IGMP packets, such as, membership report, leave group, and so on.
- In a Cross-VC environment, connecting two Edge Services Gateways with multicast to the same universal TLS is not supported.
- A single tier of Edge Services Gateway is supported.
- Single distributed logical router, that is, only one downlink per ESG.
- In NSX 6.4.5 or later, multicast is supported on a maximum of two uplink interfaces and one downlink interface per ESG. However, if an NSX Edge is at 6.4.4 or earlier, multicast is supported on a single uplink interface and a single downlink interface per ESG.
- On a DLR, multicast is supported on a single uplink interface and on multiple internal interfaces.
- Starting in NSX 6.4.7, PIM is supported on one GRE tunnel per ESG. PIM can be enabled either on a maximum of two uplink interfaces of the ESG or one GRE tunnel interface, but not on both simultaneously. To reach the sources, receivers, and RP outside the NSX network, static routes must be configured with the IP address of the GRE tunnel endpoint as the next hop.
- Active-standby high availability is supported by enabling ESG HA. Active-active high availability using ECMP is not supported.
- High availability failover time of 30 seconds.
- Cold standby, no sync of mroutes or mFIB.
- L2 bridging is not supported on a logical switch with multicast routing.
- Hardware VTEP gateways (ToR gateways) are not supported on a logical switch with multicast routing.