Rstp convergence times

[chaitanya780]

Hi Experts,

I want to understand regarding RSTP convergence times. Need some reference form standard/ benchmark to know estimated convergence times. In all posts/websites i was seeing some random values only.. please share any reference to understand. 

[balaji.bandi]

have you checked with this :

https://www.cisco.com/c/en/us/support/docs/lan-switching/spanning-tree-protocol/24062-146.html

[MHM Cisco World]

why you interest about the coverage time?
did you think that this will make you select one STP mode than other ?

MHM

[chaitanya780]

I was asking for the convergence time considering all rstp nodes.
We have some clarity in stp based on the forward delay and other params
but what rstp standard says about convergence time.

[Joseph W. Doherty]

Did you read my reply to your similar question in https://community.cisco.com/t5/switching/convergence-time-mstp-rstp/m-p/5134676#M566167 ?

If so, did you understand it?

The STP protocols do not predict specific convergence times as many variables are involved, some depending on what happening in the network at any point of time.

Typically, again, rapid STP variants convergence time is measured in multiple seconds while non-rapid variants convergence time is measured in some fraction of a minute.  Believe this is also mentioned in @balaji.bandi reference.

BTW, IMO, any STP variant is too slow to support hitless network performance.  If that's a goal, you need to use faster converging technology.  However, STP still useful to try to preclude accidental L2 loops.

[Ramblin Tech]

"... but what rstp standard says about convergence time."

I did a quick scan of IEEE 802.1D-2004 (802.1w RSTP was incorporated into .1D with 2004 edition) and did not see an absolute max convergence time requirement in the standard. Clause 17.1(c) does state as a protocol design requirement that:

The active topology will, with a high probability, stabilize within a short, known bounded interval,
minimizing the time for which the service is unavailable for communication between any pair of end
stations

So for a given topology, it should be possible to put a bound on the convergence time, but that same bound would not apply to all topologies.

Clause 17.32 Table 17-5 goes on to limit the amount of time that an individual node can spend processing a received BPDU before it must transmit a BPDU on all ports and cease to learn or forward frames as an absolute max of 1.0 seconds. Furthermore, Table 17-5 provides for an absolute max of 0.2 secs between the expiry of the helloWhen timer and transmission of a BPDU on all ports.

It has been a long time since I looked closely at STP and could very well be wrong here (I am open to correction), but the way I interpret this is that once a bridge's hello timers have expired with a neighbor, indicating a topology event, that bridge has 0.2s to send BPDUs containing the TCN. Each receiving bridge then has 1.0s to propagate the BPDU, which continues throughout the reach of the spanning tree. The bound for the topology therefore is set by the size of the topology.

So how big can a topology be? I believe the size of the topology is limited by the maxAge value for messages originating from the root bridge, which has a max value of 40sec per Table 17-1. If the root is a the center of the topology then the topology could have "radius" of 40sec for propagating BPDUs, with each bridge hop allowed an absolute max of 1.0sec to process and propagate a BPDU. The topology could then be 80sec (or 80 hops at 1s/hop) "wide".

Again, my STP knowledge is quite rusty, so if my understanding is incorrect I welcome corrections. The ratified IEEE 802 standards (ie, not the drafts) are freely available and I would encourage all network engineers to try to plow through one at least once just for the experience. Also note that the standard is 20 years old and network reconvergence bounds measured in seconds or tens of seconds sound absurdly long today.

[MHM Cisco World]

You get many timer when you search because cisco not use ieee standard it use cisco standard which is called rstp but in real it rapid pvst 

So any doc. That list ieee 802.1w forget it and see cisco doc. Only 

• STP—Defined in IEEE 802.1D, this is the original standard that provided a loop-free topology in a network with redundant links. Also called Common Spanning Tree (CST), it assumed one spanning-tree instance for the entire bridged network, regardless of the number of VLANs.

• Per-VLAN Spanning Tree (PVST+)—PVST+ is a Cisco enhancement of STP that provides a separate 802.1D spanning-tree instance for each VLAN configured in the network.

• Rapid Spanning Tree Protocol (RSTP)—RSTP is defined in IEEE 802.1w. It is an evolution of STP that provides faster convergence than STP.

• Rapid Per-VLAN Spanning Tree (Rapid PVST+)—Rapid PVST+ is a Cisco enhancement of RSTP that uses PVST+ and provides a separate instance of 802.1w for each VLAN.

• Multiple Spanning Tree Protocol (MSTP)—MSTP, defined in IEEE 802.1s, maps multiple VLANs into the same spanning-tree instance. The Cisco implementation of MSTP is often referred to as Multiple Spanning Tree (MST).

MHM