Hi
Say, for example, I am deploying a large amount of VLANs with a Nexus vPC toology. So I would have no spanning tree loops. Logically my topology looks like a single Core switch to Access switch link. However, ideally I would like to only allow specific VLANs on each vPC.
I understand the MST recommendation that for an MST region I should allow all VLANs in that region over all trunk links involved in the region to avoid accidental traffic black holes.
However I don't really want to allow VLANs on a link to an edge switch when there are no ports in that VLAN on the edge switch. So since I have no STP loops could I safely allow VLANs 1 to 100 on vPC 100, VLANs 101 to 200 on vPC 200 and so on but still have all VLANs in one MST instance ? MST would still calculate vPC 100 and 200 as forwarding so my traffic still gets to where it needs to. Only if I introduce a loop is there a potential problem.
I don't want to rely on VTP pruning and I can't create many MST instances and easily allocate vlans to each one because any VLAN could be required on any vPC.
So can I create 1 MST instance, map all VLANs to it and still safely remove specific VLANs from vPC's on a vPC by vPC basis ?
My alternative is to have my 1000 VLANs each running a rapid instance on each port they are active on when I really don't need all that overhead with no loops in the network.
I suppose in a non-Nexus environment this is exactly the same as a VSS Core type deployment with dual homed Access switches using MEC. No loops. Am I able to then ignore the stipulation that any trunks in an MST region should carry all VLANs in that region ?
Thanks, Stuart.
