Top 20 Practical Cabling Tips for Network Engineers – Part One

20 Practical Cabling Tips for Network Engineers

I think that most Network Engineers have forgotten or don’t know about network cabling. Things that I have learned, the hard way, in twenty years of looking after networks.

I find that most engineers think that cable is simply plug and play. And, mostly, that is true. In recent years, Ethernet standards and manufacturing have developed into reliable and “good enough” but the realities are still in place. Cabling needs some care and attention to detail to be reliable.

The Forgotten Things.

1. Cable Length Matters

The EIA/TIA standard says a 100 metres is the maximum cable length. THIS INCLUDES PATCH LEADS.  If you are using 10 metre patch leads at each end your horizontal cable run should be only 80 metres. If you have 20 metre patch leads then you can only have 60 metres in the wall.

2. Long Patch Leads Can be a Bad Idea

You can go longer – it might work. It’s more likely to work if you have good quality cable and connectors and a well executed cabling system since the electrical signal will degrade less in that environment. Cheap installs are more likely to have problems.

That’s why long patch leads can be a bad idea in some places in your office work just fine. And in other places they don’t. A 150 metre cable run probably won’t work, or worse, won’t work reliably.

Physical Considerations

3. Overloaded Cable Trays

Cable trays should not be overloaded. Suspended cable trays are mounted to something – either ceiling mounts or support from a rack underneath. If it’s too heavy, they can fall off the wall/ceiling etc. On to expensive things. That break.

4. Leave Space to Remove Unused Cables

Too many cables is not only a safety problem, but leads to poor operational practices when people stop removing cables from trays because it’s too hard or fear of disturbing cables.

5. Crushed Cables in overloaded cable trays.

Cables at the bottom of cable tray can be crushed by weight of cables. This especially applies to Cat6 cabling which is thicker and heavier than previous copper cabling. Cables which have been crushed will degrade signal propagation.

6. Hanging Cables will Stretch

Hanging cables within the rack means that means that gravity will induce physical stress on the copper core which will stretch and distort over time. A longer patch lead is heavier and will cause more shear stress on the cable. This will create signal degradation over time and leads to intermittent failures over time. Patch management isn’t just to keep your rack tidy, it has a mechanical purpose too.

7. Don’t pull too hard

Pulling cables can damage them by stressing the copper core. Stressing the copper core can cause stretching and thinning of the copper wire which affects the signal performance. In extreme cases it will cause unwinding of the twists in the sheath. That’s why standards specify 110Newtons as maximum draw force to be applied during installation. The cabling manufacturer should ensure that their cables are able to withstand this stress (cheap cabling is often not tested or designed, it’s just “made”).

Your cabling installer should use a force meter when pulling cable to Category 6 standards to ensure that pulling strength does not exceed to the stated maximum.

8. Electrical Conduits can’t be used for data cabling

Electrical conduits and termination boxes don’t work for Category 6 cabling because the right angle bend radii are smaller than the permitted bend radius for Cat 6 copper cable. Bend damage increases changes of cable kinking, copper core stress and leads to signal degradation.

Cheap Cables

9. Less Copper

Cheap cable has lower than standard diameter copper cores which causes signal loss and results in imperfect signal propagation. When non-standard copper cores are mated to standard cores it creates more interference because impedance mismatch causing signal reflections due to impedance mismatch.

Narrow diameter cores are also physically weaker and are more likely to snap or stretch further over time thus making them more likely to fail.

10. Cheap Patch Leads

Don’t use cheap patch cords. Again, cheaper cables often don’t quite match the standard and can cause signal degradation. Other possibilities to reduce manufacturing costs include:

  • non pure copper cores,
  • poor quality control over twist ratios,
  • poor packaging
  • sub standard sheathing

These things can all result in damaged copper cables.

More to Come…..

You can find the Second Part at  Top 20 Practical Cabling Tips  for Network Engineers – Part 2

Other Posts in A Series On The Same Topic

  1. Top 20 Practical Cabling Tips for Network Engineers - Part Two (31st July 2012)
  2. Top 20 Practical Cabling Tips for Network Engineers - Part One (19th July 2012)
  • Howard Marks

    More significant than the stretch on the conductors because of hanging cables is the strain on the connectors from the weight. I’ve seen several failures at the connector pin-conductor interface. This is common with self crimped cables.

    • Etherealmind

      Good point. Should have thought of that myself.

  • Julien Goodwin

    One to add *never* crimp your own cables at >= 1gb, sure it will probably work, but it’s highly likely to be worse then a prebuilt one and wouldn’t pass a certification test.

    Stocking a variety of lengths for intra-rack, desk-to-device and using fibre elsewhere (in DC’s) should be standard these days and if you consider the cost of your time when making a cable it’s likely much cheaper.

    Also if a cable is at all questionable (or the clip breaks) *chuck it*, but only after either tying it in a hard knot (competent network people won’t touch it) or cutting it in half (sane people won’t touch it). Again the cost of a replacement cable is far less then the time you’ll spend debugging it when it just barely slips out of the socket on a server hidden away in a rack.

    • Norgs

      2 knots and then snip the middle.

  • javelin

    Thoughts on using mjr21 to do inter rack cabling?
    If you are having to pull multiple runs its quite simple, but lack of demand means lead times arent in the same timeframe as pulling them yourself

    • Etherealmind

      In general terms, I don’t believe in copper cabling any more. It’s a technological dead end. The technical performance of MJR21 is only Cat5 AFAIK and not Cat6 so not suitable for 10GBaseT (well, I wouldn’t use it).

  • Will

    Could you please add your recommended vendors?

    • Etherealmind

      What ? For cabling ? I don’t get involved in vendor selection or have much knowledge of the cable manufacturers so I couldn’t make any recommendations. There are some differences between them but I’m not much bothered.

      Only recommendation: don’t buy proprietary cabling systems.

  • sganpat

    Don’t call it 20 tips, give 10 and then say more later. Call it 10 Tips and then you can have another post with more.

    Great tips by the way. Half of these I’ve never considered.

  • Engineers_R_alwaysright


    Nice Read. I would add as being network Engineer m,an years myself…
    Use ONLY AMP precision RJ connectors. It is OK to crimp your own as long as you have a
    Betts or other high quality crimper… Also
    In addition to AMP connectors only, Use the correct kind for stranded or solid wire respectively.

    One argument i always have to make is that 110 blocks and krone blocks are superior
    to any 66 block cross-connects. Its like pulling hair trying to get a technician to understand
    micro-connections at HIGHER frequency and data rates. They wiil fight you toothe and nail… Ignorance is bliss.


    • Etherealmind

      I disagree. I believe it’s completely impractical to make Cat6 grade patch leads by hand. You really need the jigs to make them, and testing equipment to validate them.

  • Paul Walton

    The overtightning of cable ties will also crush the cable. There are several reports by the IEEE to show how throughput is reduced when a cable tie is over tightened even by hand. Please don’t use the tie setting tools!

  • Michael Kantowski

    I would add the following:

    1) Stock (or plan for) many different patch cable lengths. You should have 1/2 foot, 1,2,3,4,5,6,7,8 foot cables for TOR installations. It’s OK to have a couple of cables that are 1 or 2 feet too long/short, but when you have many in the cable management that are off by a foot, it’s a problem.

    2) If you are working with fiber, carry a cleaning kit at a minimum, and if you want to ensure your work is done well, also carry a scope so you can inspect the end of the ferrule and core surface for oil and dirt.

    3) For copper, color makes a big difference. Use a different color per purpose.

    Regarding copper for 10G – DAC cables are the way to go for low latency and low cost.

  • network cabling world

    There is no way length should be an issue as all jobs these days require us to submit test results. All cable runs need to be under 295′. When cable tray is overloaded you can also have overheating of the cables. Something else that can happen is the cables at the bottom of the tray can become crushed especially when the tray drops down to the rack from overhead.

  • jamassica

    I have some basic information of this Cable. Like:

    4-pair unshielded twisted pair (UTP) cable, 24 AWG Copper Clad Aluminum, Verified Compliant With EIA/TIA /ETL, Low Smoke, CMP Plenum rated jacket.

    Can anyone suggest me more informations about this ethernet cable?