Does link speed affect STP diameter consideration?

[Stephen Berk]

I'm designing a metro fiber ring (1000 mbps) that will have a 9-10 hop diameter. IEEE says no more than 7 hops, but a colleague says that was recommended with 10 mbps links in mind. Any thoughts on STP failing if I have the extra hops on the gig ring? I'm using rapid spanning tree now with default timers. The ring isn't complete yet, but an average rt ping is under 10ms from end to end and shouldn't increase much based on the number of hops left to install and distance to cover. I don't want to redesign this if the gig links will handle the current setup. Any insight, suggestions, best practices is greatly appreciated.

[robertblasey]

Hi Stephen,

the 7 hop "limit" gets your problem if links are flapping and your timer are configured with its default values. If you have more than 7 hops, tune your timers if using traditional STP. The formula was 7hops x 2sec-Hello < Holdtimer 15sec - you had to increase your Hold-Timers or decrease your Hellos.

(or go with etherchannel or a layer-3 design) Bandwidth is only of concern to the metric and the decision of STP wich link to block.

In your case of RSTP - the 10 hops schould not be a problem. But not because of a "10mbps recommendation"

Regards

Robert

[ovt]

There are two factors that can limit STP diameter: max hops and the delay.

To the best of my knowledge 802.1d STP uses Message Age field in the BPDU for both of them: STP root announces BPDU with Message Age = 0, each intermediate switch adds 1 (hop), then, when a switch stores BPDU, Message Age is incremented each second, until it is updated by the new BPDU or until Max Age timer is reached (20 by default). The well-known max recommended diameter 7 was calculated to account for possible BPDU loss (3 BPDUs may be lost and everything will still work well).

MST uses different mechanism. It has another field Max Hops which is decremented at each hop starting with "spanning-tree mst max-hops hop-count" (20 by default) until 0 is reached.

RSTP, so far as I know, uses Message Age as hop count. This field is incremented until Max Age is reached. It doesn't add the number of seconds to this as 802.1d does. If 3 BPDUs are lost the STP topology change is initiated.

For 802.1d see this document for details:

http://www.cisco.com/en/US/tech/tk389/tk621/technologies_tech_note09186a0080094954.shtml

802.1w RSTP: http://www.cisco.com/en/US/tech/tk389/tk621/technologies_white_paper09186a0080094cfa.shtml

Overall it should be safe to have RSTP diameter 9-10.