RSTP clarification

[cisco_lad2004]

Dear all

I have gone thru the attched:

http://www.cisco.com/en/US/tech/tk389/tk621/technologies_white_paper09186a0080094cfa.shtml

What I am not clear about is the difference between the 2 scenarios:

1-Adding a new link which requires syncng the ports and pushing the blocking downstream. This does not rely on timers and therefore is faster. However, I assume total convergence would equal number of switches X BPDUs hello timer.

2-Topology change, which means a port going from blocking to forwarding. Convergence equals TC while timer X 2 hellos.

On secanrio 1, when a port is moved from blocked to forward, should process in scenario 2 be invoqued ?

Any good links are welcome, meanwhile I keep looking

TIA

Sam

[Francois Tallet]

Hi Sam,

-1- The hello timer is not involved because there is a kind of "flash update" to use L3 terminology. As soon as there is new information to be propagated, a BPDU is immediately sent out (there is a rate limiting of the BPDUs in case too many changes occur). So the convergence will basically depend on how fast a BPDU can be propagated to the edge of the newtork...

-2- The sync does not trigger a topology change. So a port temporary blocked by a sync will not generate a topology change when unblocking. However, the change in the network that initially triggered the sync is likely to have already started a TC.

Regards,

Francois

[cisco_lad2004]

Hi Francois

Thanks for ur useful comments.

If I understand u well, an outage situation behaves the same way as a new link addition. AT the end of the day a sync very likely ends up in a port transiting from blocking to forwarding and therefore BPDUs with TC will be flooded.

In both cases (outgage or new link), the sequence should be as follows:

1-Sync negotaitions.

2-Switch react by putting a port to forward state.

3-TC state, BPDU flood during TC while.

Do u know of any good documents on the subject, cisco PDFs are a bit thin.

Regards

Sam