It seems such a simple concept, cable management. In practice, it’s sometimes rather difficult to enforce.
Take, for example, the state of the cables under a floor in a datacenter that will remain nameless. The engineer’s task was trivial: add one more cable on a server to segment iSCSI traffic.
“I’ll just pull this tile… and route it under the floor…” she thought.
|How not to route cables|
“Oh, god!” she screamed. “What the?!?” Her plans for the day just took a detour.
I think she’ll need to schedule a downtime to unweave cables and reroute them so they can be managed.
I’d say this is a case where
- there were way too many people involved
- people involved were doing too many different things
- there were no defined policies or procedures
- no oversight or enforcement
If you think those network cables are bad, take a look at those power cables. They’re buried under that nest, too. Judging by the state of the network cables, I’d bet they also regularly had problems with circuits being overloaded and either tripping or, worse, melting and possibly catching fire.
My belief on datacenters has long been: if anything is done clearly outside of sound engineering practices then it should be disconnected and removed immediately — no questions asked — then re-engineered properly. Sure, there are some grey areas. There are some not so grey areas. But, generally speaking, that’s my view.
A much wiser engineer than me once said to me about datacenters, “The only way to keep it manageable is to be an asshole.”