rstp convergence on shared link

[sarahr202]

Hi every body!

Here is what my book says:

" RSTP does not do anything differently from stp on link-type shared links."

" RSTP does not improve convergence over link-typed shared connections"

===================================

But lost of bpdu is detected in 6 seconds ( 3*hello) unlike 20 seconds in stp, a port in discarding state connected to shared medium, will detect the loss of bpdu sooner(in 6 seconds). Is it not right?

======================================

Here is what my book says about rstp

" If the DP( designated port) fails, switch can quickly move the back up port from a discarding state to learning state and then forwarding state"

How long it takes before back up port transitions into forwarding state after DP fails on hared medium ?

My calculation is 15 seconds+ 6 seconds(3*hello.( for learning state) , so around 21 seconds . Which is in fact an improvement over stp where a blocked port will take 50 seconds to transition into forwading state.

If i am correct then my concept is in conflict with what the book says ( please read the excerpt).

Rstp:

An alternate port connected to shared medium, will detect loss of bpdu in 6 seconds, then spend only 15 seconds ( learning state before finally settles into forwading state.

stp:

a blocked port connected to shared medium, detects the loss of bpdu in 20 sec( max-age), then spends 30 sec( listening and learning ) before finally transitions into forwading state.

====================

Again i see an improvement in convergence over shared mdium when rstp is used.

Thanks a lot!

[Francois Tallet · Accepted solution]

You are right, the aging in 3xhello time will provide an enhancement over spanning tree. However, you still have 2xforward-delay, as both listening/learning stages still exist in RSTP on shared segments.

In fact, there is a further enhancement in the latest IEEE spec that allows going a little bit faster on shared segments, but Cisco does not implement it (it require cooperation from all the switches on the shared segment and would be difficult to enforce).

regards,

Francois

[sarahr202]

Thanks Francois.

"listening/learning stages still exist in RSTP on shared segments."

Here is what my book says:

When a port needs to be put in forwarding state from discarding state on shared medium,, it will be first put in learning state then forwarding state".

So there is no listening state. So port will spend one forward delay( learning state) before reaching forwarding state.

Have a nice weekend!

[Francois Tallet · Accepted solution]

Hi Sarah,

Discarding = Listening ;-)

The final state for a designated port is forwarding. It starts in discarding (or listening) state for fwd_delay, then learning for fwd_delay before reaching forwarding.

regards,

Francois