Tuesday, October 11, 2011

What does "pruning of leaf nodes" mean?

A friend of mine was reading draft-mahalingam-dutt-dcops-vxlan-00 - VXLAN: A Framework for Overlaying Virtualized Layer 2 Networks over Layer 3 Networks: and had a question about what the following meant: "This will enable pruning of the leaf nodes for specific multicast traffic addresses based on whether a member is available on this host using that specific multicast address."

For those not familiar I will not recap VXLAN since I only quickly skimmed the IETF draft, plus there are lots of acronyms I have yet to learn.

To answer my friend's question:
If I understand correctly, VXLAN extends layer2 traffic over layer3. In order to send broadcast traffic VXLAN uses IP Multicast. A mapping of broadcast to multicast groups is established. Hosts (hypervisors) send upstream notifications to routers/switches notifying them of join/leave requests to the multicast groups. The hosts send these requests when VMs are available/not available on the host.

The IETF draft says there will be management of the IP multicast group by the switch/router via a management channel, notified by the VTEP (hypervisor).