2

I have a client who has been experiencing a significant amount of performance issues in regards to their telemarketing floor. On my initial assessment, I found that where the ethernet is punched down, the last 1-2 feet of the cable jacket is stripped leaving a massive bunch of the individual pairs all bound together with no jacket protection. Basically, the guy who installed the cable stripped the ethernet where it enters the patch panel, leaving the remaining pairs bare all the way up to the punches.

I know that the jacket primarily protects the cable from physical and environmental damage, but with this amount of exposed wiring (48 punches = 384 pairs) in such a concentrated location, I'm wondering if the exposed pairs would have any effect on network performance? And if so, how would I go about testing it?

  • 3
    Don’t waste your and your client’s time. Hire a professional cable testing company. – Ron Trunk Jul 19 '18 at 13:26
  • I ran cable tests, but tools I have only measure for shorts, cable length, and pair match. One thing that made me draw the above conclusion was that while I was toning, isolating the cable was damn near impossible which is when I discovered the last stretch of cable was all bare. I will hire a testing company if needed, but I don't want to request my client drop cash on another tech until I know if it's a reasonable concern. – mlxs Jul 19 '18 at 13:32
  • Are the switch ports showing errors? – Ron Trunk Jul 19 '18 at 13:32
  • They're unmanaged switches and I didn't have time to swap in my test switch for monitoring. I will today though so hopefully that will give me some info. I ran throughput tests in the small amount of time I did have, one between hosts on punches I did myself and one between hosts on shabby punches. Tests were about 20% lower latency on good punches, but I didn't have time to run as many as I'd prefer. – mlxs Jul 19 '18 at 13:39
  • 3
    I’ll reiterate my first comment. – Ron Trunk Jul 19 '18 at 13:52
5

What you have is an amateur cable installation; it sounds like an electrician or old-school telephone installer did the work. A professional cable installer will only strip back the jacket (which does more than simply protect the pairs and separate one cable from another) as little as necessary to make the termination. The installer is also supposed to provide the customer with the results of the test suite for the correct category of each cable installed.

There should be a 10' service loop of cable at each end of the horizontal cable that could be used to correct the situation, but I'm not sure I would trust any of the cable because exceeding the maximum pull tension (25 lbs.) or minimum bend radius (4x the cable diameter; one inch on a .25" cable, so a two-inch loop) during installation can permanently ruin a horizontal cable, which needs to be solid-core cable that is easily damaged in installation.

What the electricians and telephone installers do not understand is that modern network cables require far more than simple electrical connectivity. The answer to this question gives the primary tests that a cable must pass for the cable category.

Unfortunately, The cabling will probably need to be replaced. It can be tested (with a proper tester: very expensive, or can be rented, but a certified installer will have one) first, but I seriously doubt the cabling will pass the test suite.

  • 1
    Thanks for the informative response. I just don't have the money to drop on certification gear, but I've always stripped only about the last 1/2 to 1 inch for the patch itself, I've never seen a full panel worth of wires bound up like that. Judging by the state of their overhead cabling segments (some of which was shredded during a wall demolition), I think it's safe to assume they should (at the very least) re-patch the cabling if not replace a portion of the infrastructure. Still, I'll recommend a certification run to them as well in case they want to be sure. – mlxs Jul 19 '18 at 14:11
  • You can get an installer to perform the cable certification. At least you would know what you are working with. – Ron Maupin Jul 19 '18 at 14:26
  • Yeah, just most of the environments I work in, I'm cleaning up after some severely low-budget chop shop has installed a jacked-up piecemeal network. Most "installers" I run into are hardly installers, much less actual low-voltage technicians. – mlxs Jul 19 '18 at 14:31
  • I'm often amazed that companies that completely depend on a network will be so cheap so much on the network infrastructure (cabling, network devices, firewall, etc.). When that fails, they are out of business. It's like trying to run a marathon with a badly sprained ankle: maybe you will make it to the finish line, but probably not. – Ron Maupin Jul 19 '18 at 14:45
  • In my experience, the hardest thing I deal with on a regular basis is translating potential financial consequences of shoddy digital infrastructure to individuals in sales/home improvement industry. Majority of my jobs are restoring networks that have degraded due to a business that has expanded on top of a network that was never built to sustain real traffic. Last month I worked on a network that had a mere 10 cables split and spliced to serve data and voice to 48 different endpoints; three of the splices were patched with scotch tape and one was capped with a band-aid 0_0 – mlxs Jul 19 '18 at 14:51
2

Supplementary to the excellent answer.

If they are having "significant performance problems" due to the physical cabling plant then these will appear in the counters maintained by the ethernet controller. These are displayed by the operating system (eg, Linux's ifconfig and ethtool; Cisco's "show interfaces"; Juniper's "show interface ... extensive"). Even better all OSs and NOSs will have a way of collecting these statistics via SNMP -- see the EtherLike MIB in RFC3635. I would graph all of these variables for all of the hosts and switches using a simple collection tool like LibreNMS (there's a world of equally-fine choices). If all the EtherLike error counters are zero for a number of days then the cabling plant -- as poor as it is -- is doing the job.

This gives you a cheap way of determining if the issue for this fault is the cabling plant or another factor. Then you can plot a course from that discovery.

Once you are collecting real numbers you can also come at a solution more simply if it is the cabling plant. You can reterminate one cable and see if it sends the error counters to zero. Remember to pop the wallplate out and check the cabling practices at that end too. Unfortunately one poor cabling practice doesn't exclude other poor cabling practices elsewhere on the cable path (such as lack of enforcement of turn radius at the top of the wall).

You also need to fix your processes. Someone paid for this work without it being of a satisfactory standard. There are specifications for cabling installation for each of the categories. When you purchase a "Cat 6" installation then you are purchasing installation practices as well as the cable. Your procedures should have included an inspection and simple audit prior to payment, perhaps even test results for the highest-numbered cable per bundle.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.