Minimal Cable Lengths?

Unanswered Question
Apr 30th, 2007

My co-worker (sysadmin type) has brought up that there are "minimal" cable length requirements when running from our 3750G to the servers.

I am aware of max distances, but never heard of a min (switch to server) requirement.

Can anyone chime in on this?

It's all CAT6 cabling from 3750G to SGI servers.

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (1 ratings)
Loading.
mohammedmahmoud Mon, 04/30/2007 - 14:13

Hi,

There is nothing called a minimal cable length in any communication system.

HTH, please rate if it does help,

Mohammed Mahmoud.

barrygowin Mon, 04/30/2007 - 16:41

MM,

Thanks for your reply. I have to agree, with one exception. When calculating ETH cable you have the formula 4Xdiameter. This only applies to bending the cable, as far as I know.

I must say that my co-workers are very "forceful" in his beliefs.

I have never heard of such an issue in my CCNA/NP learnings....

Thanks again.

scottmac Mon, 04/30/2007 - 17:27

As is the usual, "It depends."

From live component to live component, the *recommended* minimum is one meter, with an "absolute" minimum of 18" (~1/2 meter) @ 10/100 speeds.

The issue deals with Attenuation to Crosstalk Ratios (ACR) and the feedback loop intrinsic to all ecectrical connections.

Note that in the above, I said active to active .... transmitter to receiver. In cases where there are jumpers from a resource to a panel, and client to a panel and the overall distances are greater then a meter, there is -usually- no issue with minimum jumper length (i.e., panel-to-panel cross-connects).

The problems will appear more in full-duplex connections, becasue that provides the mechanism that would have pulses on one pair of the cable creating crosstalk on the/another pair of the cable. With half-duplex, only one pair would be active at a time.

At gigabit speeds, since all pair are active in both directions concurrently, it is also especially susceptable to crosstalk in a "too short" scenario.

Your best method to check whether this is happening on your network would be to use an in-line "sniffer" / analyzer and look for CRC errors, runt packets, late collisions (or any collisions in a full duplex environment).

Good Luck

Scott

barrygowin Mon, 04/30/2007 - 17:41

Scott,

Much appreciated.

All that said. I should be clear, either way, running a 3ft patch cable?

I will monitor a bit more closely for the CRC's and such, but not seeing any noticeable for the last 3 months with many 3 footers in place.

Again, appreciate the input.

scottmac Mon, 04/30/2007 - 19:14

Yessir. A three foot / one meter cable should be fine.

It's always a good idea to occasionally monitor your port stats for errors though, things like the cleaning crew sucking up your jumpers in the vacuum cleaner, Sally Sue Secretary rolling over the cable, and other "hazards" (to your infrastructure) always present and looking to take down your network ;-)

Good Luck

Scott

Danilo Dy Tue, 05/01/2007 - 06:36

Hi,

There is a minimum lenght for each type of cable because too short a cable will encounter crosstalk specially for higher speed.

For UTP copper cabling its always been overlook because anybody with a crimper can create a UTP patch cable. But for fibre, only certified fibre cabling technician can create one patch cable or factory manufactured, they have a strict compliance for the minimum length for fibre patch cable.

I'll get the fibre patch cable minimum length table and post it here, I remember I have them in my laptop in the office :)

Actions

This Discussion