STP and RSTP interoperability

santo.suy · ‎10-01-2003

In a switched network where part of the switches are configured with Rapid PVST+ and part are with PVST+ (migration from STP to RSTP case),

when a network failure occurs, I assume that part of the network converve slowly because of use of STP and the other part converges quickly because of the use of RSTP.

How do you explain that there is never a loop in the network while part of the network ports change its state more quickly than other ?

I have understood that with STP and MSTP interoperability, this issue is resolved by the concept of virtual bridge (STP sees MST region as a single switch). But if we use RSTP there is no such thing as virtual bridge.

A possible response is that with STP/RSTP, the switches configured with RTSP force a slow convergence in the same way as STP in case STP BPDU is seen on the network ?

milan.kulik · ‎10-01-2003

Good question!

I haven't found anything about it on CCO.

IMHO, the best migration procedure would be to start migration with a contiguous RPVST+ cloud containing the root switch for all VLANs. This should work similar way as MST region - changes inside this cloud should have no impact to the PVST+ rest of the network and vice versa.

Regards,

Milan

jwesthoff · ‎10-21-2003

Hello Santo

I have found on CCO a document that says something about convergence when interoperate with 802.1d Legacy bridge.

"It is important to note that 802.1w inherent fast convergence benefits are lost when interating with legacy bridges" this is found in the doc, "Understanding Rapid Spanning-Tree protocol 802.1w"

I hobe that will help you

Regards

Jack Westhoff

santo.suy · ‎10-21-2003

Thank you, Jack for your research.

I know already this paper, but this does not explain what is the behaviour of the whole switched network when a part use STP and the other part use RSTP and is quicker to converge.

Any way, if you have other information, I would be glad to know them.