Cables: to label or not to label....that is the flamewar

[Roy S. Rapoport]

On Mon, 26 Aug 2002, Dave wrote:
> We're having an, ahem, "discussion" with an internal customer (lab user)
> over whether or not my group (sysadmins) should be labeling patch cables
> in her (software QA) lab.  We've found that as often as not, the users
> of these very loosely-controlled software development and QA labs move
> any cables we've labeled to other equipment without re-labeling, and
> it sends us (and them) on wild goose chases when troubleshooting.
>
> We've proposed a compromise of putting a unique cable serial number
> on both ends of each (new) patch cable.
>
> What do y'all do w.r.t. labeling in uncontrolled lab areas?  (What do
> you do in access-controlled datacenters, for that matter?)

I tend to work on the basis of the assumption that bad information is worse
to have than no information -- I'd rather not know something than think I
know something and have it be wrong.

Based on that, we've only labeled patch cables in environments where we
expected to have decent control over their usage.

We have clients with remarkably static instatllations, where labeling a
cable with the port and machine name on both sides has worked well for us,
though in these circumstances we've found generally that the amount of time
we spend tracing cables is proportionally lower anyway; in all other
situations, we've attempted to come up with a good mix of colors for cables
(leaving aside two standards -- red for console and orange or yellow for
crossover) so it'll be easier to trace them.

I like the idea of unique serial numbers; it might be difficult to do with
reliability unless your cable intake process is much more regimented than
others I've seen.

In general, I'd also say that:
A) One group should be responsible for cabling;
B) That group should do whatever it wants about labeling.

-roy