CAT5/5e/6 Cable Performance

[radioraiders]

On my home network I have all 1000Base-T equipment (cards + switches), but using CAT5e cables and getting thruput about 400Mbps. A while back I bought some Belkin CAT6 cables assuming the perfomance would improve, but to my suprise it got worse ... 200Mbps So I cursed Belkin for selling me sh!t cables and put my 5e cables back.

Today I was going to order some more cables and noticed there were STP and UTP options Looking into it, I realized STP=Sheilded Twisted pair (and U=Unsheileded). I never realized this option existed with ethernet cables, I assumed they were all sheilded. So, I'm guessing the old CAT6 cables I bought were UTP (?). Would it be correct to say then that STP CAT5e cables perform better than CAT6 UTP cables then?

...and is there a big difference in cable manufacturers? Before I assumed Belkin was one of the better manufacturers, but since the 200Mbps CAT6 cables I bought from them, I'm not so sure...

[streaker69]

On my home network I have all 1000Base-T equipment (cards + switches), but using CAT5e cables and getting thruput about 400Mbps. A while back I bought some Belkin CAT6 cables assuming the perfomance would improve, but to my suprise it got worse ... 200Mbps So I cursed Belkin for selling me sh!t cables and put my 5e cables back.

Today I was going to order some more cables and noticed there were STP and UTP options Looking into it, I realized STP=Sheilded Twisted pair (and U=Unsheileded). I never realized this option existed with ethernet cables, I assumed they were all sheilded. So, I'm guessing the old CAT6 cables I bought were UTP (?). Would it be correct to say then that STP CAT5e cables perform better than CAT6 UTP cables then?

...and is there a big difference in cable manufacturers? Before I assumed Belkin was one of the better manufacturers, but since the 200Mbps CAT6 cables I bought from them, I'm not so sure...

CAT5e and CAT6 inherently have 'some' shielding in the way the wires are twisted together. There is STP cable, but it is mostly used in areas were interference from either RF or EMI sources are present. I've installed it once in a police department where the 150W radios of the police cruisers were interfering with the network.

You cannot install STP cable unless the rest of the equipment you're using is going to properly ground the cable. So you'd need a patch panel that is going to ground all the connectors that are plugged in. STP cable is much harder to work with as well as the shielding makes the cable very stiff. You also need to purchase special RJ45's for STP to allow for a shielded connection.

One more thing to add:

When you're running network cable in any environment you need to be aware of where you're running it. You should keep it at least 3feet from florescent fixtures (the ballast especially) and you should always cross 120v/240v lines at right angles. CAT5e/6 should never be run parallel to power lines. When terminating, you should untwist the cable no more than 1/2inch. I just ran some line in one of our stations and had to keep it 6 feet from the 480V 3phase VFD's to prevent interference. I did not run STP, but just kept aware of the potential issues so I could use UTP.

[Thorn]

Network speed can be also be affected by things like the radius of the cable bends around corners. Don't be sure it was 'sh1t cables', as there are many things in the environment that may cause slower speeds.

[streaker69]

Network speed can be also be affected by things like the diameter of the cable bends around corners. Don't be sure it was 'sh1t cables', as there are many things in the enviroment that may cause slower speeds.

this is also true. I can't tell you the number of times I've seen CAT5 stretched or pulled around tight/sharp corners. I've noticed that the CAT5 I've been using lately, the outer insulation actually changes color some if it's pulled too hard. It goes from a darker blue to a lighter blue.

[radioraiders]

Don't be sure it was 'sh1t cables', as there are many things in the enviroment that may cause slower speeds.

I'm an engineer, I know about variables That's why I tested the cables (right out of the box) sitting at my desk with the switch and 2 computers (all 1000Base-T). I simply swapped the CAT5e cables out and the CAT6 cables in, leaving everything else identical (even transferred the same files). And the result was significantly lower thruput. I'm having trouble explaining/understanding this. Hence, the sh!t cables remark... or the thought the CAT6 maybe UTP (?) ...or some other possible explanation...?

[Virchanza]

What do you mean by "reduction in throughput"? Do you mean that if you try to copy a file over TCP that you can only get 200 Mbps where you used to get 400 Mbps? If so, does the reduction in throughput come about as a result of re-sent packets because previous packets were in error?

[Thorn]

I'm an engineer, I know about variables That's why I tested the cables (right out of the box) sitting at my desk with the switch and 2 computers (all 1000Base-T). I simply swapped the CAT5e cables out and the CAT6 cables in, leaving everything else identical (even transferred the same files). And the result was significantly lower thruput. I'm having trouble explaining/understanding this. Hence, the sh!t cables remark... or the thought the CAT6 maybe UTP (?) ...or some other possible explanation...?

Fair enough, I mentioned that because a lot of people don't understand that environmental factors can influence the throughput speed. There are installation guides for cable, and reading them may help you determine what issues may be causing the speed drop. Streaker69 mentioned a couple, like the fluorescent ballast, but there are quite a few others. One that drive me crazy in a couple of clients' offices, is wrapping the power, network, video, etc., cables all into one bundle. Trying to explain crosstalk and signal attenuation to someone who can't program a VCR's clock, and wants everything to look neat, can be an exercise in patience.

[radioraiders]

There are installation guides for cable, and reading them may help you determine what issues may be causing the speed drop.

I don't remember the cables coming with any instructions, I assumed ethernet cables were "plug-and-play"...but maybe not so simple afterall?

What do you mean by "reduction in throughput"? Do you mean that if you try to copy a file over TCP that you can only get 200 Mbps where you used to get 400 Mbps? If so, does the reduction in throughput come about as a result of re-sent packets because previous packets were in error?

I didn't dig into it. It was 2 WinXP machines connected via a switch, I copied/pasted a DVD (VOB files) from one to another an measured the throughput with a bitrate tool. With the CAT5e transfer speed was around 400Mbps, when I removed them and put the CAT6 the best was about 200Mbps. With the CAT5e back in it was back to 400Mbps... I got pissed off, put the CAT6 cables in my closet and got on with my day...

Guess I have to read up on ethernet cables....

[Virchanza]

I didn't dig into it. It was 2 WinXP machines connected via a switch, I copied/pasted a DVD (VOB files) from one to another an measured the throughput with a bitrate tool.

What you did was copy a file over TCP, most likely using the Samba file sharing protocol.

TCP is a "connectionful" protocol that sends back acknowledgments when data has been received without error. Machine 1 says "here's some data", Machine 2 says "yeah, got it", Machine 1 says "here's some more data", Machine 2 says "yeah, got it". Machine 1 says "here's some more data", Machine 2 says "yeah, go it". And so on.

There are two things that can screw this up:
1) Machine 2 gives no response to acknowledge it received data
2) Machine 2 can respond saying it received corrupt data

If such a screw up happens, it results in a retransmission of the data.

With the CAT5e transfer speed was around 400Mbps, when I removed them and put the CAT6 the best was about 200Mbps. With the CAT5e back in it was back to 400Mbps... I got pissed off, put the CAT6 cables in my closet and got on with my day...
Guess I have to read up on ethernet cables....

As far as I know, when you change cables, there's only two things that can effect the transfer rate:
1) Propogation delay (e.g. if you had a cable from the USA to Japan you'd notice a delay)
2) Signal degradation (e.g. attenuation and interference) causing errors that result in re-transmission of data.

Now I'm assuming that you're not running a cable from the USA to Japan, so let's rule out the first cause.

We're left with the second cause... and it sounds fishy to me. The reason it sounds fishy is that I've heard anecdotal reports of transmission error on a LAN being something like "one packet per month".

When you hook up the CAT6 cable, your throughput gets cut in half, which suggests to me that there's a lot of errors... WAY more errors than you should be getting, which suggests to me that the cable is defective.

In MSWindows, open up a terminal and type "ipconfig /all". Somewhere there it should give you statistics about errors on the LAN. Copy the DVD file again using the CAT6 cable and see what error statistics you get.

(Propagation delay can also affect the switch's ability to detect collisions, but the cable shouldn't be long enough to affect that, and also there should be no collisions if the connection is full-duplex)

[terminal86]

In MSWindows, open up a terminal and type "ipconfig /all". Somewhere there it should give you statistics about errors on the LAN. Copy the DVD file again using the CAT6 cable and see what error statistics you get.

Let me correct you , because that is wrong. ipconfig /all doesn't show any statistics.

Instead use:

net statistics (Server / Workstation)