STP vs RSTP convergence times

danb83

Am I right in thinking the following?

STP:
Hello time = 2 secs
Max Age = 20 secs
Listening = 15 secs
Learning = 15 secs

= 52 secs to convergence


RSTP:
3 missed BDPUs @ 2 sec each = 6 secs
Learning (no listening) = 15 secs

= 21 secs to convergence


Also, are there any differences between STP blocking port, and RSTP alternate port. Or is just a naming thing?

thanks

notgoing2fail

boswd1983 wrote: »

Am I right in thinking the following?

STP:
Hello time = 2 secs
Max Age = 15 secs
Listening = 15 secs
Learning = 15 secs

= 52 secs to convergence


RSTP:
3 missed BDPUs @ 2 sec each = 6 secs
Learning (no listening) = 15 secs

= 21 secs to convergence


Also, are there any differences between STP blocking port, and RSTP alternate port. Or is just a naming thing?

thanks

STP

---

The hello time is 2 seconds. The Max Age Timer is 10x the the hello timer. This is important. Because it's not always 20 seconds, it's 20 seconds because the hello timer is 2 seconds.

STP has 4 states: blocking, listening, learning and then forwarding.

Once a port is in blocking state, it stays there for 20 seconds. Then moves onto listening at 15 seconds, then learning at 15 seconds. That's where you get your 50 seconds.


RSTP

---

The max age is 3x the hello. So a max of 6 seconds. There's no blocking port in RSTP. It's discarding state. Discarding replaces blocking and listening. So you only have discarding, learning and forwarding.

DevilWAH

actuly RSTP can have a convergence time of the time it takes a BPDU to travel from the center of the network to the outside..

The main difference in RSTP is that all bridges can send BPDU, not jsut the root.

So in the instance of a switch goes down, any switch that has a link connected to it will notice (as its link has gone down) this switch will then bring up its alternitive port and send out a BPDU to nebiour switchs still alive to say it has done so. (so no timer need to time out, this happens almost instancley). this BPDU then travel through the network update all other switchs resulting in sub second fail over and covergence of the network.

The hello times do some times come in to play, but in theroy for most switch and link failers in a RSTP network, failover and down time is limited to < 1sec.

notgoing2fail

DevilWAH wrote: »

actuly RSTP can have a convergence time of the time it takes a BPDU to travel from the center of the network to the outside..

The main difference in RSTP is that all bridges can send BPDU, not jsut the root.

So in the instance of a switch goes down, any switch that has a link connected to it will notice (as its link has gone down) this switch will then bring up its alternitive port and send out a BPDU to nebiour switchs still alive to say it has done so. (so no timer need to time out, this happens almost instancley). this BPDU then travel through the network update all other switchs resulting in sub second fail over and covergence of the network.

The hello times do some times come in to play, but in theroy for most switch and link failers in a RSTP network, failover and down time is limited to < 1sec.

That's an important point. The BPDU's come from all switches with RSTP whereas with STP it come from the root down to the switches.

saravanaboobathy

According to me STP's convergence time is 30 to 50 seconds only... but in RSTP convergence time is very flow its near <1 second only ..... if u don't belive that create the new configuration in Packet Tracer Software and check it out... in that software after the configuration connection will be established (STP calculation is running over there) while running it will be orange color after the calculation it turns into green color.... here you can count the time...

danb83

But if the RSTP port learns for 15 seconds, how can convergence be less than 1 second?

cheers

DevilWAH

boswd1983 wrote: »

But if the RSTP port learns for 15 seconds, how can convergence be less than 1 second?

cheers

Do the CCNP SWITCH exam and you will learn all

the forward dealy is not normal used. RSTP is much more of an active process between all switchs in a network.

They handsake on ports so check they can bring them to a forwarding state before they bring them up (but this takes millisecionds) they also only flush mac address on the ports that have failed, so when bringin up an alternitive port, they still know all mac address on any other port they still have up. They don't need a long learning phase, as the inpact of re-learning mac address is less. (plus switchs are much faster).

But this really is a CCNP topic, if you look in to uplink fast and backbone fast in common STP. This is the kind of thing that runs network wide on RSTP.

The forward (listing / learning) delay is generaly brought in to use when RSTP meets STP. and it has to fall back to using the STP timers.