On Thu, Apr 04, 2013 at 15:26:25, marc edwards wrote:
> Subject: Re: Nexus 5K vPC Peer link 1Gb over OM2 Fiber
> 
> I would agree that vPC doesn't forward much data under normal 
> circumstances... But what about forwarding mechanism in failure scenarios?
> If not addressing these during design, then one learns the hard way if 
> the unfortunate situation happens. (ie switch failure down stream on 
> chassis one causing forwarding only on chassis 2 but also a link 
> failure on chassis two upstream to router)
> 
> The whole concept of multi chassis solutions is to provide HA in 
> failure scenarios which won't interrupt or cause severe limitations to 
> data forwarding. If not setting up the vPC peer link for such 
> scenarios, than why go with a multi chassis solution at all?
> 
Planning for failures, that's crazy talk.  The CCIE is all about designing stuff that doesn't make much sense.  I've carried that into the real world :)
If you re-read what I wrote, I wasn't defending running it at 1G, I was just saying that under normal circumstances there should be little to no traffic across it.  You have a design issue if there is.  I also mentioned that I create mine using two links, a design recommendation.  Offhand, I wasn't sure if it would form over 1G, but checked and found that it will throw an error.
I would venture to say that more problems come from Layer 3 VPC, than would arise from the choke of a 1G connection as a VPC peer link.  Just throw a management port of a Netapp in the mix and watch the fun begin.  I would like to see some of the limitations of the peer link relaxed if layer 3 is involved.  After all, I'm burning 20Gbps worth of ports on both devices for a failure that might happen once in the next 5 years.
-ryan
Blogs and organic groups at http://www.ccie.net
Received on Thu Apr 04 2013 - 19:45:37 ART
This archive was generated by hypermail 2.2.0 : Wed May 01 2013 - 06:47:40 ART