STP and RSTP congergence time

In my experience, RSTP can be from around 1 second down to less than

Bert

Thanks, The port to change stage is not depends on hello package. right? For RSTP it has alternate port and backup port. The alternate port I can think it is blocking port in STP in another switch. How about buckup port, does it need to be config manually?

The alternate and/or backup links are computed automatically. To detect a failed link, you don't use hello packets, but rather the MAC_Operational parameter, which in turn is set according to each specific LAN type. The reason failure detection can be so fast is that in modern LANs, the active or backup/alternate link should never be quiescent, even when idle, so if an alternate path is available, it can be activated very quickly.

This is in IEEE 802.1D, Clause 17, which then refers to other clauses in the same Standard.

Bert

Thanks, I am trying to understand your "words" For backup port, assume there are multiple ports on the same segement, before topology change, Does the backup port automatically random selected? after the active port down, the backup port immediately become active port. I think it shouls impact whole tree.

They show pictures in Clause 17. A backup port connects to a true redundant link between two switches. That's the easiest fix. If the two switches are still good, but the link between them goes bad, going quickly to the backup link is very simple.

Alternate links do change the spanning tree, but there are tricks you can use to prevent loops when going to an alternate link. For instance, only choosing an alternate that leads away from the root node.

Bert

Thanks Albert. =46rom all switches power up to stable stage ( assume every switch power up same time), Do STP and RSTP use same time or RSTP faster than STP? I think no difference for this stage. it has to select root first then ... RSTP faster than STP only when link failed and then recoved back case. right? TIA st

RSTP is even more efficient from powerup. RSTP assigns only three states to each switch, compared with five states for STP.

Bert

Thanks, after rstp enable, It stay in discard stage. Does it take how long go to learning stage and send/receive BPDU during the discard stage?

st

RSTP can very quickly jump from discard to forwarding, if it receives the appropriate BPDU from another switch. Two switches can propose and agree with one another, without having to wait, listen, and learn.

Bert

"RSTP can very quickly jump from discard to forwarding, if it receives the appropriate BPDU from another switch. Two switches can propose and agree with one another, without having to wait, listen, and learn."

You meant that it can happend during from power up to stable stage? before I think it only happen after link failed then go to stable stage.

st

=46rom power up, or after a configuration change, it makes no difference. RSTP does not depend on the timers timing out, as STP does. RSTP instead works from the BPDUs of neighboring switches immediately. STP is very conservative, working to prevent loops and to prevent duplicate or misordered frames. Topology updates propagate throughout the spanning tree before frames are forwarded. RSTP instead does not wait, even if this risks the potential for duplicate or misordered frames.

The timers established in RSTP are mainly for backward compatibility with STP, in case some switches in the spanning tree are only STP- capable. Other than that, RSTP can do better than the timers imply.

Bert

Thanks, make me more clearly. =46rom powerup, I still think that it should elect root bridge first. The root election should same as stp. right? st

Yes, and all switches initially assume they are the root, until they hear from other switches with lower IDs.

Bert

Thanks!