Author: Brian Warshawsky
VI. Thou shalt know what cable goes where
The only place to start with managed cabling is at the beginning. Try to never run cables directly from a switch to a rack full of servers. A patch panel between the two eliminates cable clutter effectively by having longer runs come together in one spot before running into the switch. The other benefit is that the patch panel provides an effective numbering scheme for the cabling right off the bat. I use the existing labeling on the patch panel, then use a marker to write the port number on the run back to the server. Then, that same port number is written on the end of the patch cable that plugs into the switch. Doing this allows me to instantly trace a cable across the data center without having to resort to downtime to tone out the cable or run the Fluke across it.
Another good habit to get into is to use different color cables to represent different types of devices. For instance, in my current work environment, yellow Cat-5 cables are data to servers and workstations, blue represent voice, red is for switches and routers, and green is reserved for wireless access points. Doing this allows me to make quick deductions on what cable must go where based simply on color.
Something else that might not seem as obvious at first is the importance of cable support. If you have a datacenter with a raised floor under which cable runs, when you bring the cables up to the back of a patch panel or switch, don’t let the full weight of the cable hang off the connection. Doing this can cause the ends of the wires inside the termination to slowly pull out over time, leaving you with a dead cable that runs under 30 feet of floor. You can avoid this problem by using rack-mounted cable management products. These often make cabling both more efficient and aesthetically pleasing. However, if you don’t have the luxury of a rack or cable management products, you can usually get away with strips of Velcro wrap and zip-ties, both of which are available at any hardware store.
When you’re running cable, try to think ahead. Will you need a bit of slack on the server end of the run? In my case, we try to leave between four and six feet of slack coiled and tied behind each server just in case we have to slide it out of the rack to replace a drive. Because of hot-swappable RAID arrays, we can now do the drive replacement without powering down the machine or removing it from the network.
Don’t run data cable next to power cables. Doing so can cause interference on the cable and slow or stop your data rate. The notable exception to this is shielded Cat-5 cable, which is typically only used in extreme circumstances. You should also take care not to put excess bends in the cable or to kink or pinch them. There are few things worse than finishing a hellish cable run, having everything terminated and punched down, and then finding out that you severed the orange/white wire exactly halfway across the room when the cable got snagged and kinked on something.
I can’t stress the importance of managed cabling in production environments. The cabling of a network is like its nervous system. Too many times I’ve come across closets and racks that are almost beyond hope. The problem with a bad cable setup only becomes apparent when you need to upgrade or repair something in a hurry. When a 20-minute job becomes an hour of tracing and untangling cable, most people develop an instant appreciation for cable management.
The commandments so far:
I. Thou shalt make regular and complete backups
II. Thou shalt establish absolute trust in thy servers
III. Thou shalt be the first to know when something goes down
IV. Thou shalt keep server logs on everything
V. Thou shalt document complete and effective policies and procedures
VI. Thou shalt know what cable goes where
