How is BUM traffic handled in a VXLAN fabric? Compare a multicast underlay with ingress replication.
Show the answer
Say this first
BUM means broadcast, unknown unicast and multicast. With a multicast underlay each VNI maps to a group, for example 239.1.1.10, and PIM replicates packets inside the fabric. With ingress (head-end) replication the source VTEP sends a separate unicast copy to every remote VTEP in that VNI. Ingress replication needs no multicast and, with EVPN, builds its list from type 3 routes, at the cost of N minus 1 copies on the source leaf's uplinks. Most enterprise fabrics use ingress replication.
Go deeper
Multicast underlay
- VTEPs join the group for each local VNI. The outer destination IP of BUM packets is the group.
- Needs PIM sparse mode (often anycast RP on the spines) or PIM BiDir.
- Efficient for many VTEPs and heavy BUM, but adds multicast state and a new class of problems.
interface nve1
member vni 10010
mcast-group 239.1.1.10Ingress replication
- EVPN type 3 routes tell every VTEP who else is in the VNI, and the PMSI tunnel attribute says
ingress replication. NX-OS:
ingress-replication protocol bgp. EOS builds the flood list from type 3 automatically. - Flood and learn can use a static list (
ingress-replication protocol staticwith peer addresses). - Cost example: 40 VTEPs in a VNI and 2 Mbps of broadcast means 78 Mbps leaving the source leaf. Usually small, because ARP suppression removes most broadcasts.
They will ask next
When would you choose multicast?
Very large VNIs spread over hundreds of VTEPs, heavy BUM, or when you already run tenant routed multicast and the team operates PIM well.
Traps
- Forgetting that unknown unicast is flooded too, not just broadcast.
How did you do?Book 4.3 Multicast