MST: Can a virtual bridge become the CIST root for a switching domain ?

Unanswered Question
Feb 27th, 2007

The CIST root si elected according with the CSTP rules. Each MST region is considered a virtual bridge which use STP BPDUs with a special BID. The documentation doesn't say much about it and I couldn't find anything about it, so is this possible ?

It's qquite clear for me what is happening with the IST and MSTs within the MST region but I can not uderstand the election process for the CIST (in the assumtion that a switching domain containing MSTs and CSTs was just powered up).

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
Francois Tallet Tue, 02/27/2007 - 16:02

Yes, an MST region can be the CIST root. If the bridge with the best priority is within an MST region, then the whole region will look like the root bridge for the CIST. The way it works is that you have in MST bpdus both an external cost and an internal cost. The external cost is not changed inside the MST region. So for instance, if the root is inside an MST region, an external cost of 0 is propagated within this region. The external cost is located in the CIST part of the BPDU where the root path cost is encoded. As a result, at the boundary of the MST region containing the root, neighbor bridges think they receive a BPDU with a root path cost of 0 directly from the root.

I don't have time to explain all this in detail. I've attached an excerpt of a presentation I made a while ago on MST. Let me know if it clarifies the process.

Regards,

Francois

Attachment: 
cristip Wed, 02/28/2007 - 08:26

Hi Francois

Thank you for the pdf file.

There is still one thing that I can't understand about this topic. In the situation that the CIST root is within a MST region that means the CIST operates in conjuction with the IST (the theory says IST are subtrees of the CIST) and only in this situation the elected root can be part of a MST region.

But this somehow contradicts the whole concept of a virtual bridge. In my understanding the MST should be seen by the CST as a bridge (one single piece) without the opiton for the inside bridges to participate in the root election by themselves.

Thank you

Cristian

Francois Tallet Wed, 02/28/2007 - 09:42

Hi Cristian,

The IST is using internal information. The root of the IST is the CIST Regional Root (IST master in the doc). This CIST Regional Root is a bridge at the boundary of the region, hosting the root port for the MST region seen as a virtual bridge, when the Root bridge is outside of the region. In that case, the CIST Regional Root is different from the CIST Root. Now, if the CIST root is withing the region, the CIST Regional Root is the same as the CIST Root. It just means that the IST is rooted in the CIST Root. The IST still computes an internal tree rooted in the CIST Regional Root. The IST increments the internal cost within the region. However, from an external point of view, the region is generating BPDUs with a cost of 0, just as if it was a single CIST Root bridge (because the internal information is discarded outside of the region).

Hope this helps,

Francois

cristip Wed, 02/28/2007 - 10:02

What you are mentioning above doesn't clarify the whay the CIST Root is elected.

To elect the CIST Regional Root you need first to know the CIST Root.

How is that CIST root elected ? Is it the regular CST election process ?

The logic would require the election of the CIST first and after that the regional CIST can be determined.

Electing the CIST Root using the CST election process means we don't have the MST virtual bridges celarly defined yet. Am I correct ?

Francois Tallet Wed, 02/28/2007 - 10:15

Everything is done in one shot. Documentation generally splits the root election from the computation of the topology. But in fact, both occur at the same time. You have inside the MST region an "extended" BPDU that includes both the CIST Root information and CIST Regional Root information. Initially, a bridge claims to be both CIST Root and CIST Regional Root. Then, comparing its information with the one received from its neighbors, this initial assumption is modified. So if the CIST Root is within a region (meaning it has the best bridge ID in the whole network) nobody will be able to challenge it as a CIST Root and as a CIST Regional Root either. If the CIST Root is outside of the region, everyone in the region will receive this information and accept this bridge as a CIST Root. Now, the CIST Regional Root election only occurs within the region (because again, internal information is only available inside the region). Initially, all the internal bridges assume they are Regional Root, but only one of them can keep the crown.

Regards,

Francois

Actions

This Discussion