Spanning tree vs no spanning tree

[Zeeshan Siddiqui]

All,

We have a large switch Network running RPVST with a variety of Cisco swicthes we plan to add few Dell switches 8024F running RSTP,

we want Cisco switches running RPVST to be the root bridge.

As RSTP on Dell / RPVST on Cisco do not get that well with each other we plan to disable spanning on the Dell siwtches.

My qestion here is

1> If we disable spanning tree on the Dell switch Will our Cisco switch running RPVST see the loop and block one of the uplink connecting the Dell switch. ?

Lets take a simple example  Dell-SW (no spanning tree)  has two uplink to Cisco-SW (RPVST) Dell switch will not send any BPDU as spanning tree is disabled.

2> As the cisco switch is running RPVST it will send BPDU to the DELL switch will these BPDU go in a Circle between the Dell and Cisco switch and indicate there is a loop and the Cisco switch will block one uplink ?

3> Will the same senario occur if cisco is running MSTP and Dell is running no Spanning tree ?

Any Help is much appreciated

[Zeeshan Siddiqui]

Can you please help with the above (Any help is much appreciated)

Just trying to understand how Spanning tree is going to handle the situation above

[Peter Paluch · Accepted solution]

Hello Zeeshan,

Have you tested the interoperability with Cisco's RPVST+ and Dell's RSTP? In theory, it should work because Cisco has made precautions to allow for interworking of RPVST+ and RSTP.

Deactivating RSTP on the Dell switches is not recommendable at all, and it should be used only as the last resort (and even then I would be very disapproving). Do the Dell switches support MSTP? If so, I recommend migrating your network to MSTP, although that will require a maintenance window. In multivendor environments (and not only there), MSTP is the way to go.

1> If we disable spanning tree on the Dell switch Will our Cisco  switch running RPVST see the loop and block one of the uplink connecting  the Dell switch. ?

In theory, yes, but the loop will be blocked only on the RPVST+ region boundary. If there is a loop in the RSTP region, it will exist until physically disconnected. Also, it is not certain whether disabling the RSTP on Dell switches just deactivates their RSTP support, or if it also causes all BPDUs received by these Dell switches to be dropped. In that case, Cisco switches would be unable to detect and disable the loop through the RSTP region. This fact would need to be verified on real devices.

2> As the cisco switch is running RPVST it will send BPDU to the DELL  switch will these BPDU go in a Circle between the Dell and Cisco switch  and indicate there is a loop and the Cisco switch will block one uplink  ?

If there is a physical loop in the RSTP region, and the RSTP is disabled, the BPDUs sent by Cisco switches would indeed be looped as well, forcing the Cisco switch to block the ports appropriately. However, during the switching loop, the amount of looping broadcasts and multicasts may be so strongly overwhelming on the switches' CPU that they won't be actually able to process the looped BPDUs appropriately, and they may fail to ultimately block the loop.

3> Will the same senario occur if cisco is running MSTP and Dell is running no Spanning tree ? 

Yes, it will. The problem is not in the STP version but rather in the fact that a part of the network does not run any STP at all.

Once again, if the RPVST+ and RSTP cooperation is shown to be flawed, I strongly encourage you to migrate to MSTP in your entire network. Having a switch that does not run any STP is just a very weak link in the chain, and is calling for trouble.

Best regards,

Peter

[Zeeshan Siddiqui]

Hi Peter,

Many Thanks for your reply  (much appreciated)

When we run RSTP on Dell switches and RPVST on cisco switches we have two root bridges for the same vlan Dell thinks it is the root bridge and the cisco switch thinks it is the root bridge.

I completely agree disabling spanning tree is never a good idea unfortunately It was recommended by Dell Network engineer as work around.

Moving the entire Network to MSTP is not an option considering the risk invloved.

Once again many thanks for your response I am awaiting a pair of Dell switches and Will create a LAB to test before going in production

Best  Regards,

Zeeshan

[Peter Paluch]

Hi Zeeshan,

According to what you write, the Dell switches seem to actually be running their own sort of per-VLAN RSTP that does not cooperate properly with Cisco's RPVST+. If the Dell switches ran pure RSTP in a single instance for all VLANs then the interoperation with RPVST+ should be seamless. By the way, is it possible to configure Dell switches for a pure RSTP instead of their per-VLAN RSTP? That should allow for an appropriate RPVST+/RSTP interoperation.

Regarding the MSTP and "the risk involved" - I am somewhat sad to read that, as moving to a widely accepted open standard should not be considered a risk. Somehow, the MSTP is receiving much worse reputation than it deserves. Sure, it is somewhat awkward at first, and requires a good deal of studying - but it is, after all, the STP incarnation for multivendor environments that supports VLANs nicely. If I had to risk a network-wide migration to MSTP and suffer some intermittent outages until the configuration is ironed out, I would prefer that in comparison to deactivating STP on Dell switches and wait until somebody creates a physical loop, and then causing a possible network-wide meltdown.

Best regards,

Peter