5

I am deploying a new Cisco WS-C4510R+E unit in an IDF. We have an existing C4510R+E in our main data center room (core switch), and I need to connect the new 4510 to the core one. I'd like to use the 10G interfaces on the sup's (both 4510's have Sup-7E's in them) but am guessing that I need to use a fiber pair in each sup for redundancy's sake, e.g.fiber pairs into sups since the sup's are in SSO mode.

My question is if this is an OK design, what are the possible pitfalls (say if a sup blows, how long before the link fails over) etc. I've never used the int's in a sup before to interconnect, so apologies for my ignorance...

Thanks!

2
  • Are the connections L2 or L3? your setup is fine but for me to include and answer to your question for redundancy i need to know what type of link we're dealing with Nov 27 '13 at 17:39
  • It'd be L2 - we are using end-to-end VLANs here. Nov 27 '13 at 18:30
6

My question is if this is an OK design, what are the possible pitfalls (say if a sup blows, how long before the link fails over)

This design seems fine if you don’t have any dedicated 10G uplink modules. You just need to remember that your supervisor cards will go down more than any other cards on your chassis for software upgrades. Those upgrades depend on your organization; so in most cases, this could not even be an issue.

As John Kennedy mentioned, STP with BackboneFast/UplinkFast (given your using PVST+) is a must if you want to use STP to reconverge the failure in a reasonable time-frame. This will effectively take a 50 second outage to less than 30.

I think a more elegant solution would be to put these links into an Etherchannel. Since you’re running SSO, all LACP/PAgP information will be synchronized across both supervisors in near realtime. This would also allow you to utilize a 20Gb path, as opposed to a single 10Gb with one in standby.

2
  • Was wondering if I could aggregate them in an etherchannel... I will definitely do that then. Nov 27 '13 at 21:10
  • @WillDennis That's how we have our virtual server farm set up in a StackWise configuration.
    – Ryan Foley
    Nov 27 '13 at 21:22
2

Yes, there is nothing wrong with your design and it will work.

As far as your failover is concerned:

Since your links are L2 your failover will operate as fast as you tune them to failover. I would recommend leveraging the "backbonefast" feature for your failover between the two switches.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.