I'm setting up a 100baseT network and I need to know how much the length of any given cable segment affect the speed of the pc on the end of that segment. Will I see noticeable differences in speed between computers that are 3 feet apart and computers that are 10 feet apart? 50 feet? 100 feet? 500 feet? If I really want optimal speed from my network is it worth setting the machines right next to the switch and connecting them with 6 inch patch cables? If shorter is better then is there a minimum length for a patch cat5 cable?
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations SkipVought on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
How much does CAT5 cable length affect speed?
- Thread starter kamadan
- Start date
butchrecon
MIS
You know I dont think it really matters or is really noticible. Setting your system 6 inches from the switch or 200 feet, I dont think you will notice the speed difference at all with most applications. James Collins
Systems Support Engineer
A+, MCP
email: butchrecon@skyenet.net
Please let us (Tek-tips members) know if the solutions we provide are helpful to you. Not only do they help you but they may help others.
Systems Support Engineer
A+, MCP
email: butchrecon@skyenet.net
Please let us (Tek-tips members) know if the solutions we provide are helpful to you. Not only do they help you but they may help others.
bcresswell
Technical User
You will most likely not see any difference between a 2' and a 200' run but once you surpass the 300' mark, you may realize a considerable difference. TIA standards require that any cable must be under 100m(295')from its termination point in the telco closet to its termination point in the workstation. I hope this will help you in your decision making.
- Thread starter
- #5
Thanks for the help. I have one more question. If shorter isn't better, is shorter worse? I seem to remember being told cables need to be 3 feet or so in length or they didn't have enough twists to avoid crosstalk, but I can't seem to find any minimum cable length specifications for 100baseT networks. Would using 6 inch patch cables in my wiring closet or a 1 foot crossover cable between two machines cause problems?
butchrecon
MIS
I am not sure. I woulf keep it to industry standard for minimum of 3 ft. This way you keep things ok just incase. James Collins
Systems Support Engineer
A+, MCP
email: butchrecon@skyenet.net
Please let us (Tek-tips members) know if the solutions we provide are helpful to you. Not only do they help you but they may help others.
Systems Support Engineer
A+, MCP
email: butchrecon@skyenet.net
Please let us (Tek-tips members) know if the solutions we provide are helpful to you. Not only do they help you but they may help others.
MasterRacker
New member
You can buy premade 1' cables, but I'm not sure you can use them between active devices. I use 1' cables to run from surge protectors I plug my servers into and the switch, but a surege protector is a passive device. I would agreee with James and recommend 3' or longer just to be safe.
Jeff
I haven't lost my mind - I know it's backed up on tape somewhere ....
Jeff
I haven't lost my mind - I know it's backed up on tape somewhere ....
I am also using some short (12"
patch cables between a switch and a patch panel. The difference is that the patch panel is a passive device and each port is wired to some outlet in the building, so my cable length between active devices is still considerably above 3 feet. My thinking is that the standard says min. 3 ft., so why experiment. I might use (the example of) a 1 foot crossover cable on my test bench, but never as part of a production network. I put it in the same category as non-certified media!
---Will
patch cables between a switch and a patch panel. The difference is that the patch panel is a passive device and each port is wired to some outlet in the building, so my cable length between active devices is still considerably above 3 feet. My thinking is that the standard says min. 3 ft., so why experiment. I might use (the example of) a 1 foot crossover cable on my test bench, but never as part of a production network. I put it in the same category as non-certified media!---Will
workjohnout
Technical User
Keep in mind that CAT 5 maximun length is 328 ft or 90 meters, including patch cables. Longer than that and you will need fiber, a repeater or segment the network.
Length of patch cables shouldn't matter but use a length appropriate for the distance from PC to wall plate (ie don't use 20 ft if you are only 5 ft away...).
NOW, the longer the distance, the more signal suffers. This is called ATTENUATION. However, if you use a quality tester, it will measure for these things.
Length of patch cables shouldn't matter but use a length appropriate for the distance from PC to wall plate (ie don't use 20 ft if you are only 5 ft away...).
NOW, the longer the distance, the more signal suffers. This is called ATTENUATION. However, if you use a quality tester, it will measure for these things.
- Thread starter
- #11
Just to let people know what we ended up doing, we decide to split the difference between shortest possible and the three foot reccomemdation and made our patch cables around two foot long. They appear to be functioning perfectly. Thanks for all the help!
The key problem with attenuation is that a cable over 90M is adversely affected by TCP/IP collisions. The way it was explained to me is that the ACK from a packet travels at a given speed. If the ACK is not received, the sender retransmits for a certian period. The receiving NIC may have already sent out an ACK, but due to the cable length, the ACK signal is not received before the packet is retransmitted, causing collisions. I had this problem, and alleviated it by switching the remote NIC to 10MB half-duplex. Eventually, I segemented the span with a switch (theoretically giving me another 90M), and the problem is now gone.
MWM
MIS
- Aug 9, 2001
- 200
- 0
- 0
The entire Ethernet scheme is predicated on the physics (timing) of electrical signals in cable and the 100m limit is, as Nedstar1 says, designed to be short enough to allow signaling to make the round trip. One nasty difficulty arising from a long link is late collisions.
When one looks at the level of collision and other traffic patterns it can be confusing to troubleshoot. A solid, conservative network topology spares one from troubles like this.
Dirty Harry : "A man has got to know his limitations."
Some network troubleshooters won't touch a network unless the client first fixes obvious topology and physical problems. Non-standard topologies, such as long links, too many hubs in a segment, etc., make troubleshooting futile because the resulting "network" is not really a network as such and the expected rules no longer apply.
I think this is why they say,"Go back to basics," when confusion and frustration mount.
Yours,
Mike, RCDD
When one looks at the level of collision and other traffic patterns it can be confusing to troubleshoot. A solid, conservative network topology spares one from troubles like this.
Dirty Harry : "A man has got to know his limitations."
Some network troubleshooters won't touch a network unless the client first fixes obvious topology and physical problems. Non-standard topologies, such as long links, too many hubs in a segment, etc., make troubleshooting futile because the resulting "network" is not really a network as such and the expected rules no longer apply.
I think this is why they say,"Go back to basics," when confusion and frustration mount.
Yours,
Mike, RCDD
- Status
