Interface descriptions in dynamic environments

jdevoll · ‎03-01-2007

We have 3 primary sites, site a, site b, and site c. The Network team is located at site 'a' only and each site has 100 servers. The problem is, the server group will decommission servers and re-use ports from the decommissioned server for the new server. Network Services never hears about this, so the interface description never gets updated. The server environment is fairly dynamic and I?d say that there are roughly 2 changes to the server environment a week.

What is your opinion on trying to keep track of which servers are plugged into which switch ports using interface descriptions in a dynamic environment?

Jon Marshall · ‎03-01-2007

Hi

This is really down to how good your change processes are rather than a technological one.

If servers are being decommisioned then hopefully a change would need to be logged for this which would then alert you to the change in server.

Having said that we have a change system like that and still we end up with ports labelled with the wrong server names. if your server guys sit next to your network guys you have a fighting chance otherwise it is very difficult. And you could argue that it is worse to have an incorrectly labelled port than one that isn't labelled at. I have seen the wrong servers shut down because of the label on the port.

You could tie the mac-address of the server to the port so if it was decommissioned and a new server installed they would have to contact the Network people before it would work. But this is an admin ovrehead and can create issues with things like teaming.

Overall if you can't ensure the consistency of the port label i would argue you shouldn't label ports with servernames at all.

HTH

Jon