On 3/8/2011 6:05 PM, Tegger wrote:
The Daring wrote in
:


It has to do with the varying twist of the pairs. When I run a
certification test on network cables, the separate pairs check
out at different lengths then I get an average length from the
test equipment. If you run a Cat6 Gigabit connection, all four
pairs are used for the data as opposed to only two pair when a
10/100 connection is used. Good wiring practice demands that no
cable ties are cinched real tight on any network cables or bundles
of network cables. Sharp bends in Cat5/6 can also alter the signal
capabilities of the network cables. I can test a length of network
cable with my certifier, get a good reading then make a sharp 90°
bend in the cable and it will fail the test.



I've seen many large-company installations with cable ties reefed as tight
as they'll go, and with sharp bends around corners. I've often wondered if
their employees have network problems.


Often they don't even realize it. The mechanism built into network
interfaces to correct data errors is pretty robust. So often "see it
works" is declared when something is plugged in and they can ping some
other device not realizing how crappy the throughput is because of the
amount of error correction needed.