Testing CAT5/5e/6 cable speed

HackMyBalls

Did a few Google searches with no luck. So I'm curious, anyone here know how to test cable speed to see if they're actually getting 100Mbps or 1000Mbps? Trusting the CAT version is usually good but it would be helpful to find tools to test bandwidth, latency or dropped packets on local networks. Plus I frequently have to test the speed of cables that are embedded in wall drops, so I can't exactly take a look at the CAT markings. I was thinking this device looked interesting here: $24.99 Network Cable Tester RJ45 Tone Probe Generator CAT5e CAT6 Tool Check Pinout. But I can't imagine that a hardware device is the only way to test cable speed. I know nmap and wireshark do network/port monitoring but those don't actually test line speed as far as I know. I'm sure those here who have set up VoIP networks know that if you buy phones capable of GbE speeds that it's a useless expense if you don't actually have cables capable of GbE speeds.

Thanks guys!

chmorin

Only way I know how to test stuff like that is with IP SLA's, which can be configured to monitor particular links for certain criteria to analyze things like jitter, and round trip time.

Read up here: Cisco IOS IP Service Level Agreements [IP Application Services] - Cisco Systems and here to configure: Cisco IOS IP SLAs Configuration Guide, Release 12.4 - IP SLAs--Analyzing IP Service Levels Using the ICMP Echo Operation [Cisco IOS Software Releases 12.4 Mainline] - Cisco Systems

I find it can be a good tool to monitor what your links are actually providing.

But I don't think it provides you with a "This link can do 100mbps" answer you are looking for. I generally trust the certified cable you buy that complies with the standards.

tiersten

HackMyBalls wrote: »

I was thinking this device looked interesting here: $24.99 Network Cable Tester RJ45 Tone Probe Generator CAT5e CAT6 Tool Check Pinout

That tool only checks to see if the cable is wired correctly and doesn't actually test the cable characteristics. You need a TDR and other expensive tools to comprehensively test a cable.

The best you can generally do is to generate a large amount of traffic in both directions and then monitor the error counters to see if anything is happening. You'll need to run it for a long period of time in various conditions as well. If you're getting the rated speed and there aren't any errors then your cable is fine.

As chmorin said, you generally trust that the cable is able to perform to its rated specifications. That said, just because the cable is rated for it doesn't mean that it was correctly installed which may compromise the performance. Incorrect wiring, bad crimps, untwisting the pairs too much, bad sockets/plugs, faulty hardware and interference from adjacent wiring are all potential issues.

Heero

And generally if an ethernet cable will autonegotiate to a certain speed, it shouldn't have any issues running at that speed. Additionally, you can check for frame errors on the interfaces attached to the cable.

Best bet though is an expensive testing device.

tiersten

Heero wrote: »

And generally if an ethernet cable will autonegotiate to a certain speed, it shouldn't have any issues running at that speed.

I've seen incorrectly wired cables which didn't have the correct pairing which autonegotiated a 100 FD connection but wouldn't actually work properly. It was a fairly long run and the incorrect pairing meant that the interference immunity from twisting the pairs together didn't exist.

desertmouse

You need a Fluke tool. I use an Etherscope, but there are some less expensive options. Worth every penny. Every last one.

HackMyBalls

Awesome advice guys, much appreciated. The SLA reports from an ISP will be useful, not just Cisco, but I just figured there was something out there. Etherscope sounds like the one.

HackMyBalls

Well OK, that's nuts, $2K? I was thinking a few hundred. That's highway robbery. I'll Ebay this one.

tiersten

HackMyBalls wrote: »

Well OK, that's nuts, $2K? I was thinking a few hundred. That's highway robbery.

I don't see how it is highway robbery. They're not cheap but they're really not for personal usage. Even for companies, they generally don't need one so they're not going to be selling massive number of those test tools. If you need one then you'd find a cabling company with one or a company who is willing to rent you it. $2K is nothing if you're going to be running a large amount of cable or need to diagnose issues with cables.

Some NICs now have some basic TDR abilities built into them so you can have a rough idea of where a cable is damaged.

HackMyBalls

True, but I just don't think $2K is reasonable when we're just talking about measuring link speed here, which I wouldn't call an exotic request. Seems like getting exact speeds on links actually is an exotic request though, as there really aren't any freely available programs or tools capable of doing it. Guess I'm just used to freeware and GPL software!

But I would definitely be willing to pay a few hundred, say even $500, as it would be useful for my business.

tiersten

HackMyBalls wrote: »

True, but I just don't think $2K is reasonable when we're just talking about measuring link speed here, which I wouldn't call an exotic request.

Measuring the characteristics of the line isn't simple and the etherscope is capable of doing way more than just what you want. Some of the features like TDR and toning actually require special hardware to do and you can't just run software to do it on a PC. As I said, some NICs and ethernet interfaces for switches/routers have basic TDR functionality built into them.

HackMyBalls wrote: »

Seems like getting exact speeds on links actually is an exotic request though, as there really aren't any freely available programs or tools capable of doing it. Guess I'm just used to freeware and GPL software!

Told you how to determine it in a post above. Use one of the traffic generation tools like iperf and then see if the error counters for both ends stays at 0 or starts incrementing.

If you want a tool that'll do exactly what you want and only that then install iperf and some network monitoring tools onto a cheap nettop and use that. You'll have check to see if the ethernet interface on the nettop is actually capable of saturating the link though.

docrice

While Wireshark may not test the physical characteristics of a cable (it's just software that looks at packets running over the wire, after all), you can infer negotiated link speed based on the output of the IO Graph function and looking at the amount of used bandwidth measured over time. Send a super large (multi-Gig) file over the wire and see what happens. You obviously have to account for layer 2 through 4 (and above) protocol overhead as well as source and destination disk reads / writes, but it should at least give you an idea whether you're exceeding 100 Mbps.

TheShadow

+1 on the Fluke it is pretty much the standard tool for certifying cable performance. Yes they are expensive but you should look at the price of their logic analyzers. If you get into a pointing fingers dust up it is cheap at twice the price. Most work on time domain reflection which requires tricky electronics. For the money you are talking about you are only looking at the Fluke microscanner which just tracks cable length and screwed up wiring but you are after bandwidth.

If you can deal with the bare minimum look at the Fluke CableIQ CIQ100 without all the extra bells and whistles. Amazon has them for around $950 but you will probably want the tone probe too if you deal with the typical rats nest for another $100. If you find that you are not using it as much as you thought you can always sell it later. Anything Fluke holds its value for years. lets see umm try this link
Amazon.com: Fluke CABLEIQ QUALIFICATION TESTER ( CIQ-100 ): Electronics