Slashdot Mirror
Should Network Cables Be Replaced?
Jyms writes "As technology changes, so hubs routers and switches are upgraded, but does the cabling need replacing, and if so, how often? Coax gave way to CAT 5 and CAT 5e replaced that. If you are running a 100Mbit/s network on old CAT 5, can that affect performance? Do CAT 5(e) cables get old?"
Should I have a plumber re-run copper all over my house?
I am responsible for a 17 location VPN base WAN for a retail chain. We use Cat5e for everything, but in the end, it hardly matters, because Cat3 at 10 mbps is still over four times faster than the T-1 that it talks to the outside world with.
But we don't work with large files internally, even here in the corporate office. If one is working with gigabyte sized files on a regular basis, on a local network, it would matter.
These people should be able to help you.
Seriously though, what strange question. Either the cable works and you're happy with the bandwidth it provides, or it stops working and you replace it, or you want to upgrade it. What's the complication here?
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
Do not replace network cables just to do it. That is a waste of time and money Replace them in two situations:
1) You are moving to a faster signaling speed and need better cabling. 10mbps requires Cat-3, 100mbps requires Cat-5, 1000mbps requires Cat-5e. Do not run higher speeds on lower standards, it works sometimes but often it "works" in that you get link but there's all kind of errors.
2) A cable has a fault. Sometimes they will break because of strain. In this case, you need to replace them to make them work.
Barring that, keep the cable you have. No reason to replace it just for fun. Also no reason to upgrade to new standards without a reason. It isn't as though it makes shit work better. 10mbps is 10mbps no matter if it is on Cat-3 or Cat-6. Also sometimes you get standards that aren't useful. Cat-6 is likley to never be useful for anything. 1gbps only needs Cat-5e, and 10gbps is likley to require Cat-6a. So if you upgraded a Cat-5e network to Cat-6 to prepare for faster speeds, well then you probably wasted your money and will have to upgrade again to Cat-6a if you want 10gbps.
If you install cables outside in pipe with good insulation they should last 10-15 years with few problems.
An Education is the Font of All Liberty
"If it ain't broke, don't replace it with Cat6."
Seriously, replacing cable is gigantic pain in the ass, when you could be doing better things with your time. Not to mention, it's expensive if you have a large enough installation-- this is why people are spending so much to keep Cat5e creaking along.
If it's working, and you're happy with it, keep it. If you need something faster, or it doesn't work anymore, or you need to meet new fire codes, well, that answers your question.
Remember, wires are solid state electronics. There's not much to go wrong there unless you're in extreme environments.
Well terminated cat5 cable will be sufficient for achieving 1Gb/s speeds. What's more important for maximizing your throughput is to ensure that you have your cables properly wired to support full duplex connections. In addition, all passive hubs should be eliminated and replaced with GigE switches, either managed or unmanaged depending on how much control you need.
I am becoming gerund, destroyer of verbs.
Bull - you can do Gig-E (IEEE 802.3ab) perfectly fine up to the 100 meter spec over regular old CAT-5 - http://en.wikipedia.org/wiki/Gigabit_ethernet. You don't need CAT-5e or CAT-6 unless you have incredibly shitty cable, splices, runs approaching max length, or too many patch panels along the route (IE, a crappy install in the first place).
Now, I personally use shielded CAT-6 for everything, but I believe in overkill :)
I'd love to see the manufacturer's cost of materials for mega priced "audiophile" cables like this. Do they really spend more on "high quality" materials (even if it's useless) or do they just make it out of the same stuff as regular cable and then try to keep a straight face while they take your money?
In industrial settings, cable breakage from mechanical flexing stresses isn't uncommon.
In residential or commercial use though, your typical ethernet cable shouldn't really degrade over time unless it is subjected to frequent connection. My personal experience leads me to believe cables running between patch panels and routers are pretty reliable, but those between cubicle walls and connected to laptop docking stations fail most frequently.
You can't legislate goodness. Let each to his own destiny, by will of his freely made choices.
Much like oil, these substances will never truly 'run out'. We're not shooting them out in to space, after all. What does happen is that, like those mines you mentioned, they become too expensive at today's prices. Tomorrow's prices, on the other hand (since demand will rise while the supply drops/becomes more scarce) will make these mines economical once again. Some mines shut down over very small price changes in the metal.
Copper's price, which was astronomical last summer, has dropped considerably. I know because I bought all the 12/3 cable for my house last summer as a hedge against rising prices and now it costs half what I paid. :(
And we can always harvest these materials from garbage if the price gets high enough to make it economical.
-b
No offense, but I've stopped responding to AC's.
or long term exposue to magnetic fields causes elecrical resistance and damages the switch over time.
Your post made good sense, apart from this bizarre statement. Did you just make this part up or something ?
You seem to be confused with link speeds versus transfer speeds which can be very different.
I'm sorry but 1Gbps is not equal on all hardware. Some hardware isn't even capable of going that fast, see the vast majority of consumer oriented network cards that come 10 for $1. As much as I love those realteks they are slow. Compared that with a server class NIC and you get dramatically different amounts of throughput.
Hardware matters a lot and so does cabling. Just because you aren't getting errors doesn't mean you're going as fast as possible given your wiring and in my experience good luck getting actual gigabit speeds over Cat5 or Cat5e. Cat5e can at least do it over short distances like say 20' but much past that and your performance will indeed drop. It's easy to measure. Create a ramdrive on an ftp server and put a single large file in the ramdrive. Now initiate transfer to ramdrive on the other end. I have to transfer multiple terabytes when I arrive back at HQ after events so I notice those little performance differences.
Of course since the bad old days I have moved on to fiber which is proving to be much more resilient to new technology.
Yes, transfer speeds can vary greatly depending on the hardware being using, including the Ethernet controller, bus, CPU, drivers, etc.
No, transfer speeds will NOT vary for two cables with the same 1000BASE-T link and no Ethernet transmission errors. I'd suggest you get some hardware that lets you monitor for Ethernet transmission errors (not TCP/IP errors) and run your test again.
A 100M Cat5e cable will transfer at the same rate as a 100M Cat6 cable IF there are no transmission errors. In my experience, a well terminated Cat5e cable does not get regular Ethernet transmission errors with a 1000BASE-T link.
Yes, but -- just like with CPUs -- we are *over* the hump in the S-curve now.
With a few small exceptions (HD-Video and the like), Core CPUs, a gig or 2 or RAM and GigE are *enough*.
Really. :-)
Someone at Denon should go to jail for that.
each port on the core switch can be manually set to a fixed 100mb full duplex (and ignore auto-sensing)
I just died a little inside. Fix the client, don't bandage the switch. I won't say autosensing problems don't exist, but I can't remember a time in the last decade across dozens of Cats and thousands of edge ports when duplex problems weren't caused by either stupid users forcing their duplex and thus requiring the switch to go half-duplex or wretched terminations at the jack or panel introducing frame/crc errors that would have continued anyway.