How bad is a spanning-tree loop across two sites ?

[Fab T]

Hello,

I need to connect 2 datacenters together and span some VLANs across both datacenters. I will have 2 dedicated ethernet links between both datacenters.

How bad would it be to connect switches as shown below and let spanning-tree break the loops (dashs being inter-datacenter links) ?

VLANs that need to span across datacenters are connected to access switches (using active/backup bonding, no LACP because no stack).

IC switches would be used to connect all routers together (2 distinct VLANs, OSPF on top of it and iBGP between loopbacks).

I have read this Cisco document but unfortunately I cannot use vSS. One reason is budget (I have hardware that I must use), the other reason is that we experienced a nasty outage because of stacked switches and it has been definitively ruled out.

I am unsure about EoMPLS and VPLS.

I have read horrible stories about spanning-tree going haywire and I certainly don't want that to happen ...

Can somebody provide some insight on this setup ?

[Joseph W. Doherty]

What variant of STP are you planning on using?

What's the latency and bandwidth between DCs?

[Fab T]

I plan to use RPVST+ with less than 100 VLANs. I could use MSTP but I understand it would have more management overhead (manually assigning VLANs to instance), right ?

The links between DCs will be fibers (2 distinct providers, going 2 different paths). Expected latency is less than 1 ms, bandwidth is 1 Gbit/s.

[Richard Burts]

The title of the original post asks how bad is a spanning tree loop across two sites. The answer to that question is that any spanning tree loop is very bad, no matter if it is across two sites or is in the same site.

The post then goes on to ask about a design where access switches connect to each other and to IC switches. The IC switches connect to each other, to the access switches, and form the connection between sites. As long as the trunks are correctly configured (be careful that vlans match up and that native vlans match up) and spanning tree is correctly configured then this design should work fine and spanning tree will detect and prevent loops.

HTH

Rick 

[Fab T]

I actually started asking a less specific question and then changed it without updating the title, sorry :-/

Thank you all for your answers. I will pay extra attention to uplinks interfaces.

[Joseph W. Doherty · Accepted solution]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

With rapid-STP (or MSTP), with your latencies and bandwidths, STP behavior should be much like LAN behavior.  I.e., I agree with Rick, I think you'll be okay.

Regarding rapid-STP vs. MSTP, yes there's a bit more configuration involved with MSTP, but your switches would only need to run one (or two, if you want to use 2nd link) STP topologies instances, rather than one for each VLAN.  If your "less than 100 VLANs" means nearly 100, you might still consider MSTP.