The hunt for interop

So here's what I had typed up a few years ago:

"HP was a huge disappointment.  I received four 5900 switches for testing and promptly set about working to get them to speak to my Avaya test switches using SPBM.  After days and days of work and dozens of e-mails back and forth with both HP and Avaya engineers, I have finally given up any hope of them working together.  HP's SPBM implementation is pretty fractured and limited right now, which makes it an unappealing choice for me anyway, but I'm not pinning the interop fault on them.  Here's what I've found:

Pro:

Switch clustering is simple and works pretty well
HP's CLI is pretty easy to figure out if you're used to Cisco gear, particularly with the command-alias feature
Third party optics work just fine with these switches
OSPF and static routing were both easy
Hard to beat the port density
Upcoming feature (RRPP) adds useful metro-e capabilities

Con:

I could only get link aggregation to work with LACP, no static etherchannel (minor complaint)
SPBM implementation is complicated and feature-limited
Doesn't interop with Avaya switches (not sure whose fault that is)
Switch software upgrades can cause issues with IRF"

I got to the point where I was hitting my head on the wall with this stuff, neglected to post anything, and ended up dropping this blog as a side effect of my frustration.

Come to find out, interop was possible.  There's a ticker in the IS-IS TLV for SPBM that specifies both the backbone VLANs (used for multipath/load balancing) and whether or not each one is in use.  In Avaya's case, every backbone VLAN is automagically in use.  In HP's, you need to manually assign an ISID to a BVLAN before it will be marked used.  If there's a BVLAN mismatch, the Avaya/HP interop will fail to form relationships.

The biggest disasppointment in this whole ordeal was that, rather than try to fix an issue with a platform that they supposedly supported, HP's engineers kept attempting to steer me in different directions.  They tried to steer me to both RRPP (proprietary SP-oriented fast-reconverge ring protocol) and to traditional layer 3 models, but neither of those fit my needs with any attempted semblance of simplicity.

In the end, we pulled the trigger on the Avaya SPBM purchase, and I was very happy with the result.  Regrettably, I ended up leaving my position, and the folks who replaced me decided to trash my selected solution in favor of a traditional layer 3 model.  I don't blame them, though.  They experienced a multitude of vendor hang-ups that I didn't personally experience, including having support dropped on recently purchased (<30 days) switches.  C'est la vie.

The biggest disappointment for me in this whole ordeal is that I've neglected this blog to such a significant extent.  I started this to try to act as some sort of change management platform for my professional growth and to (hopefully) allow other people to learn from some of the trials I've encountered without going through months of learning/troubleshooting on their own.  To that extent, I feel like I've failed, but I'm going to give it another go.

Hopefully you've picked up something worthwhile from this entry.  Thanks for tagging along!