The other day I was working on setting up an equipment rack in a new location. I had ordered a piece of equipment and it had arrived. You can see it at right. It made me think of a bottle of “gluten-free” seltzer water that I had encountered recently in a grocery store. Old-timers will recall the early days of rack mounting. Manufacturers of networking and computing equipment had apparently decided that anybody who was purchasing a piece of equipment and needed it to be “rack mountable” was extremely cost insensitive. A 24-port ethernet switch with four rubber feet on the bottom, for placement on a shelf, in those early days might cost $400. A rack-mountable ethernet switch, identical in absolutely all functional respects and differing only in that it possessed mounting flanges to fit an equipment rack, would cost $900.
In those days a rack-mountable panel with, say, 24 ethernet jacks, suitable for terminating 24 ethernet cables, might cost a couple of a hundred dollars.
The panel that you see above is what arrived the other day. It cost $30. You snap keystone jacks into it. This is very handy in many ways. For one thing you can mix and match, using jacks of particular colors to help keep track of the functions served by particular circuits. You can put single-mode or multi-mode optical fiber connectors into selected positions, or F9 RF coax connectors.
Yet another handy thing about this style of panel is that you can punch down a particular position later and it is not a fuss or bother. With the legacy style panel you pretty much need to punch down all 24 positions at once. If you only punch down, say, 20 positions, and later you decide you need to punch down two more positions, it is very awkward to do so because you have to flip over the entire panel. You risk pulling loose a previous connection. With this new style of panel you just punch down one more keystone jack and then snap it into the panel, without disturbing any other positions on the panel.
But back to the gluten-free seltzer.
Old-timers will recall the brouhaha in 1998 and 1999, as the year 2000 drew near. Everybody was worried about the many systems that had been programmed in a way that used only two digits to keep track of the year. When “99” incremented to be “00”, maybe the system would crash, or maybe the “carry the 1” would overflow into some other field and make trouble for some other value stored in a database. This was called “the Y2K problem.”
And indeed when January 1, 2000 arrived, there were some systems that crashed or malfunctioned, as you can read here:
- On January 1, 2000, just past midnight, the US Naval Observatory, which runs the master clock that keeps the country’s official time, gave the date on its website as “1 Jan 19100”.
- 150 Delaware Lottery racino slot machines stopped working.
- In Japan, NTT Mobile Communications Network (NTT DoCoMo), Japan’s largest cellular operator, reported on 1 January 2000, that some models of mobile telephones were deleting new messages received, rather than the older messages, as the memory filled up.
So in 1998 and 1999, one of the challenges for consumer electronics companies was to carefully check the firmware in their products to see whether their products were vulnerable in this way. A company selling, say, a data router in 1999 might say “Y2K compatible” on the product packaging as a way of reassuring the nervous would-be purchaser that the company had carefully checked the router’s firmware and had determined that the router would not crash or malfunction on January 1, 2000.
And I recall back in 1999 seeing several consumer electronics products marked “Y2K compatible” when this clearly meant nothing. One was an extremely simple ethernet hub that pays no attention to the day or date, but merely passes ethernet packets to and from all of its ports in a very simple way. The hub never learns what day it is and so has no opportunity to malfunction based upon the arrival of any particular date. Yet the manufacturer clearly felt that marking the packaging “Y2K compatible” might prompt the nervous consumer to select this ethernet hub instead of selecting some competing ethernet hub that was equally safe from any Y2K risks. The thought I guess was that the consumer might be ignorant enough not to realize that a simple ethernet hub that does not know what day it is cannot possibly crash on January 1, 2000 due to a Y2K problem.
But Tommy doesn’t know what day it is!
(Props to to the first reader to catch the cultural reference.)
Which then brings me to the gluten-free seltzer water. Lots of folks avoid gluten, although it turns out that many who avoid gluten are probably mistaken to think that it is the right thing for them to do. Yes there are some people who benefit from avoiding gluten but there are many others who just get swept up and mistakenly think that they, too, should avoid gluten just because it gets talked about so much. For the vast majority of people, gluten is actually quite good for you and is a valuable protein source just like many other valuable protein sources.
Anyway, there are some makers of packaged foods who hope to grab the attention of the nervous or inattentive consumer by marking their products “gluten free” even when it is quite clear from context that any competing product would necessarily be just as gluten-free. Seltzer water is the example that sticks in my mind, but there are many other examples of this that one can find in any grocery store but especially in health food and natural food stores.
Back to the “Cat 6” rack-mount panel.
Cat 6 means “Category 6.” In the context of computer networking, Category 6 characterizes network cabling through which one may consistently pass ethernet packets at one gigibit per second (one billion bits per second). It all started with Category 3, which was old-fashioned twisted-pair telephone wire that had been repurposed for ethernet. One could not reliably pass data faster than ten megabits per second with Cat 3. Next came Cat 5 (Category 5), which was good for 100 meg (100 megabits per second). Next came Cat 5e, and then Cat 6.
The point of Cat 5 (and then Cat 5e, and then Cat 6) was mostly that of the four pairs of wires in the cable, you twist them so that each one of the four pairs has a number of twists per inch (or per centimeter) that differs from that of each of the other one of the four pairs. The idea of course is to minimize common-mode coupling of RF energy from one pair to the next. The alert reader will recall that I pointed out in this blog article that this technique of non-identical linear spatial frequency of twists is exactly like what cicadas do to try to keep from getting eaten.
With Cat 5, you try to be pretty careful when you manufacture the cable and the jacks. With Cat 5e you try to be even more careful when you manufacture the cable and the jacks. With Cat 6 you try to be super careful when you manufacture the cable and the jacks. Of course Cat 6 cable costs much more per foot than Cat 5e, which in turn costs more than Cat 5 cable. Cat 6 jacks cost more than Cat 5e jacks, and so on.
Anyway so if you want to make an ethernet circuit that is Cat 5 compliant, or Cat 5e compliant, or Cat 6 compliant, you do the following:
- you make sure to use cable that is Cat 5 compliant, or Cat 5e compliant, or Cat 6 compliant;
- at each end of the cable you make sure to use jacks that are Cat 5 compliant, or Cat 5e compliant, or Cat 6 compliant;
- when you strip off the cable jacket to punch down at each end, you strip off only about half an inch of jacket;
- when you do the punching down at each end, you do not untwist each pair at all except to the extent minimally necessary to reach the punch-down positions;
- along the length of the cable you make very certain never to kink the cable at any point;
- along the length of the cable you make no sharp bends but only very gentle bends; and
- making very sure never to exceed the permitted 100-meter cable length.
There is test equipment that you can use to test such cable runs for Cat 5 compliance, or Cat 5e compliance, or Cat 6 compliance. The equipment has a signal injector at one end and a receiver at the other end, and it passes frequency sweeps across each pair and listens on the other pairs, looking for near-end crosstalk and far-end crosstalk. It tests for signal attenuation on each pair across a range of frequencies. And of course the equipment looks for opens and shorts and crossed pairs and split pairs. Years ago I had a Pentascanner that did this for Cat 5e compliance and I used to methodically test all of my work. It nearly always passed.
In the old days, a company that hired a contractor to do network wiring would often hire a second contractor to do acceptance testing with equipment like this, to check to see whether the first contractor had followed all of the rules and had not cut any corners.
Back to the “Cat 6” rack-mount panel. So you could use Cat 5 cable and Cat 5 jacks, and snap them into this so-called “Cat 6” panel, and the fact that it is marked “Cat 6” would do nothing whatsoever to help those little ones and zeroes pass along from one place to another. It would not magically transform the circuit into a “Cat 6” circuit.
So this is actually worse than the gluten free seltzer water. At least it is not a lie to say that the seltzer water is gluten free. It is pointless to say it, but it is not false to say it. But here, it is really false to suggest that anything about the bracket is capable of being or not being “Cat 6”.
I am going to use the panel anyway. It’s not as though there is some other less expensive panel of this same type that I got fooled into not selecting. But I will avoid allowing the “Cat 6” marking to influence anything in my activities about the panel.
One last thought about Cat 5 and Cat 5e and Cat 6.
I tend to be a cheapskate, buying the least expensive cable I can find. Yes I make sure to choose plenum rated cable if I know it is going into a plenum, but otherwise I do not pay extra for it to be Cat 6 or Cat 5e cable. Likewise I do not pay extra for my ethernet jacks to be Cat 6 or Cat 5e.
And I tend to be lazy, stripping off much more than the half inch, just so that it is easier to manipulate the wires for purposes of punching down. And untwisting more than the bare minimum, just to make it easy to manipulate the wires. And although I try very hard to avoid kinks, I am sure that every now and then I screw up and kink the cable and do not notice it. And although I try very hard not to violate the minimum bend radius requirement, I am sure that every now and then I violate the minimum bend radius requirement.
And guess what? My circuits always work at gigabit speeds anyway. And nowadays the only “acceptance testing” that I do is with a very simple little tester that checks to see that all eight wires have good connections for DC current flow.
The fact is the standard is over-engineered, and the equipment that companies sell these days is over-engineered, and the cables and connectors are over-engineered. And so I benefit from all of the engineering margins. And by far the majority of my cable runs are far shorter than the 100-meter limit and I am sure that also works in my favor.
How careful are you about the length of jack that you remove? How careful are you about the amount of untwisting that you do? Do you do frequency-sweep acceptance testing for your newly installed network wiring? Please post a comment below.