Spanning Tree in ring topology

[marco.iozzi]

Hi all. I've a 30Km ring of SMF in wich the costumer wants a network access every 2Km. Therefore, two segments have been hypothesized (8 switch segment, 4km apart and each phase shifted 2km) to cover the entire stretch of 30 km with a total of 16 2960S switches. draft included. The 2960S switches at the ends of the segment will be stacked together. Is it possible in order to protect the network from loops to utilize the spanning tree protocol in this ring topolgy? I fear that the number of hop in this ring will exced the scope of the spanning tree.

thanks

Marco

[vmiller]

Assuming you set a root bridge at one end, at least one of your orange links will be in a blocking state as a loop prevention mechanisim this is what spanning tree does. I do not know of any hop count concerns, thats generally a layer 3 metric.

Spanning Tree is generally on by default in cisco switches.

[Peter Paluch]

Hello,

Actually, there is also something similar to a  maximum hop protection mechanism in the STP. Each BPDU has a field  called Message Age. The root switch sends the BPDU with the Message Age  set to 0, and each "hop" further from the root bridge, the Message Age  is incremented by one as the BPDU is propagated further. Different STP  version then use the Message Age field differently:

• The legacy 802.1D STP switches expire the BPDU after (Max Age - Message Age) seconds,  i.e. the first non-root switch expires the BPDUs after 20 seconds, the  second switch expires the BPDUs after 19 seconds, etc. Note that on the  19th switch, the BPDUs would expire exactly after 2 seconds which is  also the Hello time, thereby making the BPDUs expire almost at the same  time they are received, and just a small skew in the timing would make  the 19th switch believe that the root bridge has ceased to exist. The  20th switch would not even be able to consistently maintain its state  because it would expire the BPDUs after 1 second although they are sent  every 2 seconds. Thus, there is a limit on the number of "hops" in the  802.1D STP.
• The 802.1w RSTP and 802.1s MSTP ignore the BPDU if its Message Age is equal or higher to the Max Age.

Have a look at this document:

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

Best regards,

Peter

[marco.iozzi]

vmiller, my doubt arose reading this thread in wich is stated that 7 hops is the maximum suggested limit for spanning to work properly

https://supportforums.cisco.com/thread/150484

[Peter Paluch · Accepted solution]

Marco,

If you'll have a look at the document I have suggested in my earlier reply here, there is an in-depth explanation of the STP timers and why they are set to the particular values of 2, 15 and 20. These values have been computed for the legacy 802.1D STP, assuming that there are seven switches at most in a row, to provide a stable STP operation. You could enlarge the STP domain beyond the limit of 7 switches, however, it would then be recommendable to tweak the timers to allow for longer propagation and processing times in a network to maintain the stability of the STP.

Please note that these timers were computed with quite a big safety boundary, and from the viewpoint of the processing speed of modern switches and their CPUs, these timers are strongly conservative. Also, the 802.1w RSTP and 802.1s MSTP do not rely on timers, instead using the Proposal/Agreement mechanism. The only limit in RSTP/MSTP networks is the Message Age which should not reach nor exceed the value of Max Age.

Best regards,

Peter

[marco.iozzi]

Ok Peter, many thanks for your aswer