Replies: 2 comments
-
Citing the NVIDIA document you linked to:
It still uses layer-3 VNI (there's no magic), it's just that it performs intra-subnet bridging instead of intra-subnet routing. It's really the question of "do we bridge or route within the egress directly-connected subnet". Looks like another blog post is in order ;) |
Beta Was this translation helpful? Give feedback.
-
The way I read that RFC, the devices that do not support intra-subnet IP forwarding can ignore those announcements, so I guess we don't have to worry about that. |
Beta Was this translation helpful? Give feedback.
-
SR Linux follows RFC9136 for ip prefix rt5 announcements, and does not support RFC9135 (which involves 2 RTs, one for l2 and a second for l3)
In FRR terms this corresponds to the
prefix-routes-only
optionConsequently, inter-subnet forwarding does not work as one would expect. How should we handle/flag this, if anything?
Beta Was this translation helpful? Give feedback.
All reactions