COBBLED CABLING - Part 2

by Wayne M. Krakau - Chicago Computer Guide, November 1995

This is the second part of yet another series that I’m doing on cabling. (Considering my last name, perhaps I should have called it "Kobbled Kabling".) I shall continue with my litany of cable madness.

The next case involves another LANtastic network using Thin Ethernet. I was visiting the site ostensibly to regain access to printers and to reconfigure their security system. While cleaning up each combined server/workstation and plain workstation - a major task in itself - I kept having workstations randomly fall off the network. (That’s technicalese for losing the connection to the file server.)

Since I had stabilized the configuration of the computers, I suspected the LAN cable. One obvious, and very common problem, was the lack of grounding. In addition to the 50-Ohm terminators on the ends of Thin Ethernet, the specification requires that one of those terminators - arbitrarily chosen - be grounded. This cable (as well as most others that I encounter) wasn’t grounded. This can slow a LAN because packets (the smallest unit of data sent) are often sent multiple times before they get through, and it can compound other errors, but is almost never a serious enough standards violation to blow workstations off the LAN. More detective work was in order.

I examined each individual length of cable and the T-connectors and found that although the cable was the proper RG-58 A/U, the BNC connectors on the ends of each length were atrocious. I suspected that they were crimped by an amateur (noting that I am only a "talented" amateur in LAN cabling, and I defer to true professionals in that field) using the unreliable, inexpensive (All right, I’ll admit it. I really mean CHEAP!) crimping tools sold by a certain national chain of electronics stores. (Notice how I deftly sidestepped mentioning the chain’s name? I don’t want to spend the rest of my life arguing legal niceties in front of one of Judge Ito’s professional kin.) This company’s crimping tools work well enough for cable TV, but not for LANs. Moreover, even their tools require at least some expertise to work correctly.

The cabling system had multiple occurrences of loose pins (the part that caps the center conductor of coaxial cable), some to the point of dangling freely, misshapen ferrules (the part that goes over the cable and gets crimped), misaligned bodies (the outer portion that actually makes the connection), and the ever popular random bits of braid wire (the crisscrossing mesh that forms the outer conductor of coaxial cable) sticking out all over the place. The terminations were so bad that I found one spot where I could slightly rotate a length of cable just short of its end and turn the network on and off like a faucet! I found out later that this spot could have been used as a sensor for their burglar alarm. If you even slightly bumped the table that supported the PC at the endpoint of this cable segment, the cable would be rotated just enough to shut down the network! That’ll make a fairly good motion detector.

As I suspected, the client told me that their "computer guy" installed the cable. This is yet another case of someone dabbling outside their field of expertise and having the client suffer for it. Considering what a mess the overall network was in, perhaps the term "amateur" could be applied to the computing aspects of this "computer guy’s" expertise, too.

In the next case, I was called in by Bruce Kahn the president of my favorite professional data cabling company, Telnet Communications Consultants, Inc. (Buffalo Grove, IL, 708-215-0003). His client was designing a new network and he wanted to work jointly with my company (KBCS) in redesigning the cable plant in conjunction with redesigning the whole network.

Another company had already laid cable throughout the building. Their contract called for cable appropriate for future use with 100-megabit networking. A friend of one of the client’s managers had informally reviewed the cable system and had noted some potential length standards violations. He was a computer professional in another area of computing, so he suggested that an outside company review and possibly even test the cable plant. That’s why Telnet was called in. Bruce also noticed that the plan for the network, in general, seemed illogical and inefficient. That’s why Bruce called me.

After the manager’s friend cast doubts on the original system, the cabler partially rewired it to split it into smaller segments. Telnet’s technicians tested the system and found that a third of the segments violated the 100-meter (about 328 feet) 10Base-T (Ethernet over Unshielded Twisted Pair - UTP - cable) standard. The lengths were more than 450 feet!

This did not even include an allowance for a 3-foot jumper cable in the wiring closet to go between the concentrator and the patch panel, or the 12-foot (average) jumper going from the wall socket to the workstation. It also didn’t include an allowance for extra terminations. The termination at the concentrator socket and the one at the network card socket are a part of the standard, but additional intermediate terminations require a distance "penalty" or correction factor. We usually use a 10-foot penalty. There were three extra terminations, the "IN" port of the jumper block, the "OUT" port of the jumper block, and the wall socket adjacent to the workstation. That extra 30 feet of penalties added to the 15 feet allowance for jumpers means that even more segments were too long!

Telnet added secondary concentrators, placing them so they could halve the length of the over-spec segments. This brought them back into agreement with the rules.

The riser cables (between floors) were 25-pair cables that were broken out into 12 data links at the patch panels of each wiring close. That’s a violation. They might have gotten a 10-megahertz signal down those cables, but a 100-megahertz signal would never make it.

Telnet reterminated the riser cables to use the 568B standard for terminations, using eight individual wires for each data link, for a total of four data links. They also cut the "leftover" wires so that nobody could inadvertently use them later, throwing the system back out of spec again.

The original cablers had strung multiple pairs of fiber optic cables to each wiring closet based on the manager’s friend’s mistaken belief that it was required for high-speed networking. They didn’t bother warning him about this bit of misinformation. For good measure, they used older, potentially less reliable, and much more expensive methods of terminating the fiber optic cable. Telnet left this cable in place in case it is ever needed to overcome the distance limitations of UTP cable.

After the client’s cable system was stabilized and retested, we looked up the details of the original system and calculated the segment lengths before the manager’s friend spotted potential problems. There were multiple segments over 900 FEET long! (Remember. We are dealing with a 328-foot standard.)

The original cabler did much of the labor needed to bring the system into compliance, and they refunded a fraction of the client’s fee. They never did reimburse the client for the outside resources needed (Telnet and KBCS) or for employee time lost during the project.

The really scary part is that this cabler is a huge multimillion-dollar telecommunications company, more known for its telephone services than its LAN expertise. (Notice how I weaseled out of an explicit name mention here, too! This company has more lawyers than Madonna has bedpost notches.) There is no excuse for them selling cabling services in a field in which they are ignorant of the standards. Also, why in the world don’t they KNOW the standards? This is a case of major-league dabbling outside your expertise.

It is important to remember that cabling standards for high-speed data (like LANs, for instance) are much more strict than those for telephone cabling. This holds for material, tools, and even the techniques used in actually laying cable. A reliable data cabling company can double as a phone cabler (especially since it’s very cost effective to string both at the same time), but the reverse is an invitation to disaster.

Please join me next month as we continue our exploration of the darkest unexplored regions of LAN cabling.