6

I can go down to my local graybar right now and purchase some AllenTel 48 Port Cat 6 Patch panels I need (part number AT66-PNL-48). They are $275 each. That seems a bit high considering I've seen other brands for far cheaper -- $165 for this brand.

Is there a considerable difference between the two? All I'm doing is running cable from my network rack to my server rack (which are about 10 ft away from each other in the same room) under an elevated floor using cat6.

Is this something i can 'cheap out' on and save a bit of money and still have the same performance? Thanks!

ItsPronounced
  • 614
  • 4
  • 18
  • 40

3 Answers3

6

The answer to your question is probably lying in the manufacturer's data sheets: Short answer is that if the gear meets the Cat6 spec's requirements it's a "Cat6 panel" - One manufacturer may just barely meet those requirements, another may meet them with a bit of a margin, and a third may have a wide margin of performance above the requirements of the Cat6 specification.

If you're curious/interested look at the data sheets for the product & compare them to the Cat6 spec - I won't say there's no difference, but in practice I think panels from any reputable manufacturer will be electrically similar enough that you don't need to worry about it & can look at price.

Personally in practice I look at other factors (cable management & panel-to-panel connection style are biggies -- How much work will it be to cable the panels up, and if I have to move them in a year or two will I wind up having to replace all my panel-to-panel trunks or can I reuse them?). Spending less time with punchdowns & crawling under floors is always worth a little price premium in my mind :-)

voretaq7
  • 79,345
  • 17
  • 128
  • 213
1

I have found that where the cost is really going to hit you is the labor to patch these to panels together (electricians are usually not cheap ;-) ). You might want to have a gander at Tyco's MRJ-21 system (see this previous SF question). I am not sure if it meets your 6 requirement -- it is Gig though. Cuts the whole labor and is really easy to move around down the road.

Kyle Brandt
  • 82,107
  • 71
  • 302
  • 444
  • I am patching them myself, I was just curious if the cost of the more expensive panel justified it. – ItsPronounced Oct 27 '10 at 16:44
  • I'm pretty sure the Tyco/AMP MRJ21 gear is only available in cat5e, but definitely worth looking at & talking to them -- they may have cat6 stuff available and just not well-advertised. – voretaq7 Oct 27 '10 at 16:51
0

Copper is copper, gold is gold. That's the expensive part of any interconnect.

What makes a good high speed cable and connector is the insulation quality (material) and tolerance control (thickness). This applies to the PCB at both ends as well. There's some cost adder here.

Then comes the transmitter and receiver circuit. The accuracy of those chips to provide the exact impedance matching per spec, and correct impedance during the bit transition (eye measurement). This is temperature dependent, and silicon process dependent, so the Si mfr needs to put auto-tuning or calibration in place to compensate. Hitting 1 GHz is somewhat easy with any telephone twisted pair. 10 GHz is a real pain.

I have just yesterday swapped two 50' lengths of CAT5E @GBE between two NAS boxes, and I get no more transmission errors. That proves that the receiver makes a difference.

On a similar topic, we tested super-expensive Video RCA Monster Cable vs cheap General Electric cable, using a 35 GHz Tek Network Analyzer, and found that the Monster cable didn't even make 35 MHz bandwidth. The GE made 1000 MHz clean. Bandwidth required: >7 MHz, preferably 35 MHz (for analog signal).

Some issues also come from new features like the "green" switches, which change amplitude (drive strength) depending on cable length and connect speed. Sometimes they get jammed at the wrong level, and the data rate gets stuck at 100Mbits.

The ultimate test is physical measurement, which isn't something most IT centers bother with. Or a controlled breakout box which would attenuate (disrupt) the signal on some path, and check for margin. I haven't seen such a box yet.