[BBLISA] Rent/Borrow Fluke or similar for Cat6 horizontal testing

[Edward Ned Harvey (bblisa4)]

> From: bblisa-bounces at bblisa.org [mailto:bblisa-bounces at bblisa.org] On
> Behalf Of Jon Young

Glad things worked out for you.  You're right, I'm not just implying, but directly stating the inverse.  I have many times, at many companies, unionized and otherwise, new offices and change orders within existing space, found that after contractors complete all their testing, some data drops are still faulty.  Same result that you experienced.  

I always watch them work a little bit while they're working, I observe what equipment they're using and what methods they're using.  I'm satisfied on all counts.  They always have these $1100 flukes, or equivalent.  They're not using the $35 cheap-o testers.

I have never bought or borrowed the equipment to test it myself.  I use ethernet at both ends.  (I prefer to attach a laptop to a switch in the closet, and then patch a whole bunch of jacks into the switch, so I can simply carry a single laptop around the space to perform my testing.)  I find problems the contractors missed.  I call them up and complain and ask them to re-test cables in question.  They do send somebody out at their own cost, to remediate the problem.  Because they care about their reputation and repeat business.  Every contract I've ever seen includes a 30 or 45 day guarantee.  I'm an IT contractor myself, so I see a lot of new buildouts, and bring a substantial amount of business to these guys.  (But not multi $B buildouts.)

After they re-work those drops, the problem goes away, which validates my original complaint.  So far, there's not a single instance where I've falsely identified a faulty drop.  So far, it sounds like your Ethernet track record is the same as mine.  Your ethernets were failing to auto negotiate, which gave you cause for suspicion.  You got some good equipment and confirmed the fault.  As you said though, it's certainly possible some drops don't meet spec, and the NICs at each end compensate and conceal the problem.  But if it passes my test, the worst I might experience is premature NIC burnout.  Maybe it happens sometimes, I don't know, but if so, it's below my radar and passes my test criteria.

Neither the contractor, nor I, extensively re-validate all the drops after their initial sign-off in a new space.  They sign off, I normally test like 5% of the space (all pass).  The main value-add of my testing is to confirm proper labeling and locating jacks on the map.  We move into the space, some (but normally very few) of the users complain about connectivity problems.  We workaround the users' problems by using a different wall jack.  (Always buildout a little over your predicted needs.)  It's normally around 1% failure rate (like 1-2 drops in every 100 will exhibit some sort of problems).  This is low enough that we don't bother with an extensive re-validation.  Just call in about jacks # 27, 72, and 137.  Move on.  

There was only one situation where the contractor had a faulty stripper, and in a cubicle block of 50, there were like 15-20 problem lines.  That is the only situation where I did any extensive systematic testing of all the jacks, and significant random sampling of the other cubicle blocks / offices, after the contractor signed off.  I called them, they grumbled, disputed my results, because they tested them all before signing off.  I got them to agree as follows:  In the following block of 50, please test the following 12, which are clearly defective.  After you find they are in fact defective, you'll realize I have no good reason for faith in the rest, so please re-test the rest.  If these 12 drops are *not* all defective as I claim, then don't re-test the rest, because clearly I'm wrong.  They agreed to that, confirmed my findings, and re-tested and corrected everything, to at least the normal level of expectation.  If there's a 1% failure rate, it's low enough to be below the noise, and not worth pursuing.