Kitz Forum
Computers & Hardware => Networking => Topic started by: skyeci on March 19, 2022, 05:35:25 AM
-
Whilst prepping and thinking about my fttp install some years ago I hard wired our house with cat5e. I know cat6 has more bandwidth but would we as home users really see any difference in replacing all the current 5e. It was a bit of a task at the time so I am not sure if I wanted to even do it at the moment unless we really would see any noticeable difference?.... all good quality switches/patch leads etc being used at the moment.
fttp will be 1gb service.
Thanks.
-
If its done well it should handle 2.5Gbit no problems and may even run 5Gbit or 10Gbit depending on the quality. The point of 6a is merely that its guaranteed to support those.
-
Unless the cable or install are actually bad, not just reasonable, 5G should be fine across the kinds of distances we home users cable across.
10G over sub-30 metre spans may be doable.
-
Domestic distances are usually much lower than the 100 metres that these cables are certified to support at the appropriate speeds.
I've pushed my network in the other direction, using 4-core telephone cable up to about 25m, it works fine for TV streaming and many other functions and is much easier to route.
I "think" it's only supposed to run at 10mbps though I have my suspicions that I might get more than that. :angel:
Edit: In fact it seems to reach 100mbps only requires 2 pairs, it's Gigabit that needs all 4. :graduate:
It's a long time since I ran the cables and I've forgotten my researches at the time, but I think all devices are configured at 100mbps and works fine.
The cable is "sort of" (but probably not exactly) Cat3 I believe.
-
I think all devices are configured at 100mbps and works fine.
Just checked my router and all the local LAN is 100mbps. :cool:
-
I'm running 10Gbps across my CAT5e, can't remember the lengths of the cable, but I've even joined the two office cables at the patch panel downstairs, and tested between the two office PCs. I think that's about 25 meters + patch cables, and it worked fine.
-
Yes, I have 10G running over short 5e runs. Not a fan of it so don't have much but it's there where there were no other options. Rather use twinax and optical.
-
Thanks for replies. I would like to think I did a good job back when I installed it. Basic wiring tester to confirm the wiring but is there any software that can test the throughput somehow of each cable run end to end etc ..
-
Sadly not. You need the radios either side to generate the signals.
Just as with DSL it's all about attenuation and crosstalk so you need hardware capable of transmitting and receiving the frequencies you want to measure.
Doesn't use the exact same systems as DSL but doesn't use baseband either. It goes up to 500 MHz and uses 16 possible states for each Hz so needs a decent SNR along with a cable thick enough to not lose the higher attenuation 500 MHz.
-
There's plenty of software to measure the performance, Iperf is one, but not so good on Windows, there's another I've used, but can't remember what that's called at the moment.
Remembered it https://totusoft.com/lanspeed and https://totusoft.com/lanspeedserver
-
Think he's after what the physical layer is capable of given the thread is about wiring. Software isn't going to tell you a run of cable is capable of 40G when you've 1G ports either side, need hardware for that.
-
Unfortunately you need expensive hardware for that, if you want actual trustworthy measurements.
-
It’s difficult to get measurements because the result will depend on the physical layer protocols in use, so it will be more about how the link is being used than the properties of the link itself. Depends on what you want in practice though. The former might be more relevant and useful.
-
Kinda, though people will be using Ethernet so the that layer is well known. Cable qualification measures attenuation at a variety of frequencies and crosstalk between pairs. It's all a matter of SNR fundamentally. Ethernet has certain requirements across the entire spectrum and if the cable can't meet those at the specified transmit power game over.
No-one is using a variation of DSL on their 4 pair twisted pair to maximise throughput and there's no prospect of it. Cheaper to use better cabling than spend all that money on transceivers. It gets that expensive and the requirement that pressing copper is replaced with fibre.
-
Think he's after what the physical layer is capable of given the thread is about wiring. Software isn't going to tell you a run of cable is capable of 40G when you've 1G ports either side, need hardware for that.
He said "is there any software that can test the throughput somehow of each cable run end to end", surely as you've pointed out if he wanted to know the maximum speed the cable could support he would of asked about hardware, the OP also infers he's only interested in 1Gbps speeds.
-
Pretty sure a cable run is hardware, Sir. Ethernet doesn't rate adapt once connected so no point in measuring throughput between two devices directly connected to the cable run: you look at speed, duplex and errors and get that from switches.
An error free, stable cable run at 10G will give 10G as long as the equipment either side isn't a bottleneck. Software can't measure the performance of the cable run as it has to run on a computer. You're measuring performance from one computer to another via all the equipment and software layers in between, not the cable run.
Switches that are non-blocking and the error rates either side are fine. If they're very low or zero and the link is stable the throughput on the cable is what the link says on the tin.
-
Whatever, sir, we're both guessing at what Skieci meant until they come back and clarify. Yes you are correct cables are hardware, but he still asked if there was any software that can test through put :P
-
Thanks for the comments. As I don't have access to a certified tester like a fluke or something I just wondered if it was possible to see if my cabling was any good using some software to test its performance...
-
Sadly no. Any software checker will be restricted by the hardware either side and will never give a result higher than the maximum the hardware will support.
To check if a cable can support x Gbps you need hardware either side that can actually push that many Gbps through the cable.
-
I would think that without special kit then UDP ramping up to flat out transmission rate would be the best you can do on the cheap. Of course a tester that could inject arbitrary waveforms would be far better, far more stressful, but perhaps not sufficiently ‘relevant’ to your everyday world - what’s the word I need here?
-
- what’s the word I need here?
"usage". :)
-
I would think that without special kit then UDP ramping up to flat out transmission rate would be the best you can do on the cheap.
You'd need to compare it to a bad cable though, as even over fibre I get "some" retries from TCP traffic if I try to max out the link with perf3 so I'd expect UDP to always show some dropped traffic. There's just so many places things can go wrong in the software stack that it makes knowing what is normal and what is a defect really tricky.
The best you can do really is bring up the interface statistics and look if there have been any dropped packets, collisions or overruns. But again, I see dropped packets even on 10Gbit when stress testing. (or maybe I DO have a fault?)
-
Thanks for the comments. As I don't have access to a certified tester like a fluke or something I just wondered if it was possible to see if my cabling was any good using some software to test its performance...
Can you clarify whether you want to know the maximum speed your cable will support, or that it will work fine at 1,2.5,5, or 10Gbps?
-
The best you can do really is bring up the interface statistics and look if there have been any dropped packets, collisions or overruns. But again, I see dropped packets even on 10Gbit when stress testing. (or maybe I DO have a fault?)
Overruns are between kernel and NIC, collisions can't happen on full duplex connections. Drops it depends which they are however filling a link you will see drops when you actually fill it. Have to go above the link's capacity to find that capacity with software.
-
Thanks for the comments. As I don't have access to a certified tester like a fluke or something I just wondered if it was possible to see if my cabling was any good using some software to test its performance...
You are going to need hardware, but what's your budget?
Mikrotik have some good kit, Zyxel have a switch that might be perfect. You'll likely only need a couple of ports capable of more than 1G on each switch so there are options.
-
Overruns are between kernel and NIC, collisions can't happen on full duplex connections. Drops it depends which they are however filling a link you will see drops when you actually fill it. Have to go above the link's capacity to find that capacity with software.
Thanks, I was hoping someone would clarify what they mean as its something I've never really thought much about since the 10BASE2 days. It just struck me as odd as I never saw drops on Gigabit but have since being on 10Gbit.