MST region with new nexus 5548

[stephen.ehlas]

Hi

A nice MST (multiple Spanning Tree) question for you all. I have a core switch, configured as follows:

Region 0  - no vlans - priority 81992 - Root for CIST

Region 1 - all vlans - priority 8193 - Root for MST1

As part of a business migration from this core, to a new Nexus 5548UP core I want to connect them together with a trunk. The Nexus does not need STP par say, but best practice is to have it on, so its is configured as follows:

Region 0 - no vlans - priority 32768 - Root for CIST

Region1 - all vlands - priority 24577 - Root for MST1

So here's the questions:

If I attach the nexus to the current core, I assume that the current core will stay as root for region 0 and region 1 due to its lower pri and the nexus will simply become another switch in the STP domain?

Now bearing in mind this is part of a project to remove the current core completly, would i be better to configure the Nexus with Region 2, so it remains as the root bridge for its own region? My thinking here is that when I pull the plug on the current core, the Nexus will already be root bridge for its own region 2 and there will be no election tantrums

Which way is best? or is there a "third way"

regards

Steve

[Peter Paluch · Accepted solution]

Hi Steve,

First of all, I have a feeling you are confusing the terms instance and region. You are probably talking here about instances, not about regions (a region is a set of switches running the same MSTP configuration name, revision and VLAN-to-instance mapping).

If I attach the nexus to the current core, I assume that the current  core will stay as root for region 0 and region 1 due to its lower pri  and the nexus will simply become another switch in the STP domain?

You are correct. The added switch has numerically higher priorities for both instances and it will not take over the existing root.

Now bearing in mind this is part of a project to remove the current core  completly, would i be better to configure the Nexus with Region 2, so  it remains as the root bridge for its own region? My thinking here is  that when I pull the plug on the current core, the Nexus will already be  root bridge for its own region 2 and there will be no election tantrums

Recall we are talking about instances, not about regions. Even if you made the new switch to be the root for instance 2, pulling out the existing switch would cause the network to lose the root bridge for instances 0 and 1, causing a transient outage until they reconverge on the added switch as the new root.

I would personally suggest first adding the new switch, then modifying the priorities in instances 0 and 1 so that it becomes the new root, and then pulling the plug on the old switch. There will be a transient outage when modifying the priorities but I do not expect it to take longer than 500-1000ms. In fact, my feeling is that the outage will be shorter if you increase the priorities on the old switch, rather than lowering the priorities on the new switch.

Best regards,

Peter