Solved: rapid spanning tree

carl_townshend · ‎09-10-2010

Hi

when reading about rapid spanning tree, the hello's are 2 seconds, and 6 seconds for max age, so this means normally that if there was a root change, it would take at least 6 seconds for reconvergence, however under normal circumstances, it has nearly been instant if I unplug a redudant link!

Is this cause rstp has the alternate port concept? secondly, why is it instant and not at least 6 seconds as per the book?

cheers

francisco_1 · ‎09-10-2010

RSTP convergence times typically take less than 10 seconds. In some cases, they can be as low as 1 to 2 seconds depending on numbers of factors.

Hello time: The time root waits before sending periodic Hello BPDU that are forwarded by the other switches. By default, it is 2 seconds.

In rapid stp, a bridge sends bpdu every 2 seconds by default with its current information even if it does not receive any from the root bridge. In pvst, only the root bridge sends bpdu

Max Age: The time any switch should wait before trying to change the STP topology after unhearing Hello BPDU. Maxage timer does not exist in RSTP. RSTP takes an active role in bringing the network into convergence rather than passively waiting for timers to expire.

To help speed up convergence from direct link failure, RSTP has natively implemented the same type of mechanisms involved in UplinkFast and BackboneFast. when aSwitch loses its RP, it immediately transitions the Alternate port into forwarding mode. In legacy STP mode, a direct failure of this type would have taken 30 seconds. Unlike the UplinkFast mechanism, the RSTP mechanism does not use dummy multicast generation to flush the CAM entries. TCs generated by RSTP to the upstream switch clear the appropriate CAM entries associated with the broken link.

When a TC bit is set, the switch starts a TC While timer equal to 4 seconds (2 * hello interval) for all its non-edge ports. It flushes the MAC addresses that were associated with that port. The upstream switch that received the TC BPDU will flush its MAC addresses from all ports except the port that received the BPDU. This process streamlines the convergence process. In legacy STP, the TCNs first needed to be propagated to the Root, which afterward generated configuration BPDUs that were propagated back to the spanning-tree domain. The amount of time it took to converge the network was contingent upon how big the spanning-tree domain was. In RSTP, the TCs are flooded quickly to non-edge ports and RPs, and the upstream switches flush their CAM entries, resulting in faster convergence time.

Francisco.

View solution in original post

francisco_1 · ‎09-10-2010

RSTP convergence times typically take less than 10 seconds. In some cases, they can be as low as 1 to 2 seconds depending on numbers of factors.

Hello time: The time root waits before sending periodic Hello BPDU that are forwarded by the other switches. By default, it is 2 seconds.

In rapid stp, a bridge sends bpdu every 2 seconds by default with its current information even if it does not receive any from the root bridge. In pvst, only the root bridge sends bpdu

Max Age: The time any switch should wait before trying to change the STP topology after unhearing Hello BPDU. Maxage timer does not exist in RSTP. RSTP takes an active role in bringing the network into convergence rather than passively waiting for timers to expire.

To help speed up convergence from direct link failure, RSTP has natively implemented the same type of mechanisms involved in UplinkFast and BackboneFast. when aSwitch loses its RP, it immediately transitions the Alternate port into forwarding mode. In legacy STP mode, a direct failure of this type would have taken 30 seconds. Unlike the UplinkFast mechanism, the RSTP mechanism does not use dummy multicast generation to flush the CAM entries. TCs generated by RSTP to the upstream switch clear the appropriate CAM entries associated with the broken link.

When a TC bit is set, the switch starts a TC While timer equal to 4 seconds (2 * hello interval) for all its non-edge ports. It flushes the MAC addresses that were associated with that port. The upstream switch that received the TC BPDU will flush its MAC addresses from all ports except the port that received the BPDU. This process streamlines the convergence process. In legacy STP, the TCNs first needed to be propagated to the Root, which afterward generated configuration BPDUs that were propagated back to the spanning-tree domain. The amount of time it took to converge the network was contingent upon how big the spanning-tree domain was. In RSTP, the TCs are flooded quickly to non-edge ports and RPs, and the upstream switches flush their CAM entries, resulting in faster convergence time.

Francisco.

carl_townshend · ‎09-10-2010

hi there

thanks for the response, im sure I read in the cisco book that the max age was 6 seconds.... is this wrong then, so rstp has no max age ?

cheers

francisco_1 · ‎09-10-2010

It does have max age timer but i dont think max-age plays much in RSTP like pvst. The point i was trying to make is rstp react to convergence quickly rather than passively waiting for timers to expire like pvst.

Does that make sense?