You don't need to be an 'investor' to invest in Singletrack: 6 days left: 95% of target - Find out more
I've moved my Deco mesh units about and whilst doing it ran a network cable through the loft between 2 of them.
And couldn't work out why that one wouldn't show as having anything connected to it. After turning off one of the others I could make my phone connect to it and did a speed test of the internet - only 80Mbps. Unplugging the cable and it was showing 220+Mbs and the things I'd expect to connect to that unit slowly moved over to it.
Is it possible for a cable to work but be slow?
I might try putting new plugs on but otherwise I'll have to run a new cable or just leave it wireless, which is the easiest option.
Yes.
There are 4 pairs of wires in a network cable. 100Mbit/s uses 2, 1000Mbit/s uses all 4. If it's badly crimped, it could drop to the lower rate.
Also if there's electrical interference, or the cable is too long (and/or crap), you can get a high number of error packets, which could appear like a slow connection.
Make sure it’s CAT5e or CAT6, and that the terminations are T568A at each end to get the wire pairs correct and reject noise. A poor cable will definitely run slow!
Is it definitely using the network cable as ethernet backaul? You can check on the app's network map and, if not, there's troubleshooting suggestions here.
Yes... it's digital data so if packets of data arrive at the receiving end, incomplete or corrupt, the same data gets sent again, and again, and again, and again until the data is complete.
It's called packet loss, and manefests as slower speeds, as you're basically sending the same data over and over again until it's correct.
The cause could be anything.. poor quality cable, poor plug/termination, exterior electro-magnetic interference, or even a severe kink in the cable, (ooh-er missus! 🤭 😉 )
That's one of the downsides to CAT6..it's quite stiff and chunky compared to CAT5, and doesn't take well to being bent at sharp/right angles.
I’ve experienced this with new pre-terminated cables on occasion!
the most common cause is bad termination.
I’ve also experienced it with cables that once were fast and then slowed. The most common causes for this seemed to be contact corrosion, rodent damage, kinking/crushing, and fkw.
while cables are best and fastest (when they work) if your wireless backhaul is sufficient and you won’t be driven mad by the thought that a cable would make it better then you could settle for that.
A rudimentary packet loss test would be plugging in something like a laptop to the Deco access point (via an known good Ethernet cable) and running ping -t at a command prompt (using the IP address of your default gateway as the destination, ipconfig will show you that). You shouldn't see any dropped packets (and latency should be sub 5ms). No packet loss but latency spikes might indicate something going on with the access point instead (although in that case if it's not degrading bandwidth for WiFi connected devices as well that's a bit odd)
I don't know as I have much further to add. Good work, all.
If you've hand-crimped the cable and you're not used to doing the job then it's not just 'possible' but indeed highly likely that you've made a pig's bladder of it. I don't mean that disrespectfully, I've turned out many garbage terminations myself before I'd had sufficient practice to get it right most times. Even now I still occasionally think "well, that's a load of shit" and end up with a cable a couple of inches shorter than I started with. Half-decent tools help massively also, but that's a spendy solution for a single cable run.
As I've said on AV discussion threads, digital signals either work or they don't. As matty says though, Ethernet is a bit different in that it will hurl data down the cable but will retry sending data if nonsense comes out the other end. Multiple dropped packets == slower transmission speeds and there are several potential causes for this.
I once did a site visit for "slow network" issues, turned out that they'd taken a regular indoor cable and slung it out of the window and over the roof rather than make the effort to route it internally. It was pissed wet through and rotten but still worked, albeit barely. The error log counter on the router was like the stopwatch for a 100m sprint, the end digits were a blur. That got terminated and not by a crimp tool.
Have you got a cable tester? A cheap nasty tester won't give you any idea of the quality of the connection, but will at least tell you that there is one. As Aidy said, Gigabit Ethernet requires all 8 wires, but 10/100 only requires 4.
As Aidy said, Gigabit Ethernet requires all 8 wires, but 10/100 only requires 4.
So is 80Mbps likely to be running at 10/100?
It’s entirely possible my crimping is gash.
I’ve made up a few cables to run around my house, this is the first one that appears to have had a problem.
I’ll put new ends on and see if it improves.
I wasn’t sure if a network cable either worked or didn’t.
No I don’t have a tester, probably should get one.
So is 80Mbps likely to be running at 10/100?
Wholly plausible, sounds about right for "Fast" Ethernet after overheads.
I’ll put new ends on and see if it improves.
I would. It'll cost you pence and minutes.
I wasn’t sure if a network cable either worked or didn’t.
No I don’t have a tester, probably should get one.
A cheap tester is basically a continuity tester, it will tell you that you have an electrical connection between Pin 1 at one end and Pin 1 at the other. For the sake of like ten quid it's a no-brainer if you're doing your own cable runs.
An expensive tester will give you a raft of other information, but is more expensive. I think mine ran to something like £300 and it's been invaluable, but I got some bugger else to pay for it. 😁 For professional testing equipment you can add a zero onto the end of that.
I'd definitely try re-crimping in the first instance. Do one end at a time.
For longer/more awkward runs, I normally terminate things with a socket these days. Feels a bit more "pro" and harder to get wrong. Easier to swap out known good cables to diagnose things, or put longer patch cables in if you ever need to move your kit, too.
Something like this?
That looks to be as good / crap as any at that price point. It's not a Fluke but it'll tell you if wire 7 isn't connected.
For longer/more awkward runs, I normally terminate things with a socket these days.
... is a good shout. Punching down wires at the back of a socket is way more difficult to cock up than hand-crimping cables.
Sorted.
Plugged my laptop in to the suspect Deco unit with it running un-cabled and got speeds of 260ish.
Plugged the cable in between the Deco and another known good one and speeds dropped to 80ish.
Re-made the plug on one end of the cable between them, plugged it back in and now have speeds with it connected via a cable of 260+.
So looks like it was a dodgy cable. Cheers all.
I'll keep an eye on it over the next few days to make sure Deco doesn't kick things off the unit I'd expect them to be connected to like it had been.
No idea of my actual LAN speeds? Don't know how to test that.
No idea of my actual LAN speeds? Don't know how to test that.
If thats speedtest.net or similar, then that's your connection speed including your LAN speed...
If you log into your router you can see your houses actual connection speed to the outside world.
So say thats 80mbps for arguments sake...
If you then do a speed test in a web browser on a PC or from a mobile phone connected to the router, etc, and it's only 70mbps, you know you are loosing speed on the connection between the router and the connected PC/phone/tablet/whatever.
Which is kinda expected on wi-fi rather than cabled.
For example..my PC is connected to my router via wi-fi, and sees 75.1mbps on average which is kind of expected.
If thats speedtest.net or similar, then that's your connection speed including your LAN speed...
Yes it’s from Speedtest.net
I get about 260Mbps when I run it which matches my Virgin 250Mbps service.
But my LAN network should be quicker, presumably up to 1000Mbps. Not that I ever really do anything over it really, apart from backup my laptop to a drive attached to my desktop.
To measure your LAN speed, basically copy a file across the local network, divide how big the file is by how long it takes. Some file copy tools will just tell you the speed of the transfer.
I’ll give that a try sometimes just out of curiosity.
To measure your LAN speed,
Honestly, does it matter?
The vast vast majority of most people's network traffic will be Internet and unless you've got FTTP or some other high-speed solution then that's going to be your bottleneck. How often are you copying files around your network? Most folk probably wouldn't even know how, they'd use USB pendrives.
"Only" 80Mbps is still pretty quick. I've managed corporate infrastructures and the only things plugged into the Gigabit switches are servers and other networking devices, 100Mbps "Fast Ethernet" to the desktop is typical. It's not uncommon to serve just one network connection to a desk and daisy-chain the PC connection off a phone which will only pass through 100Mbps regardless of what's at the other end.
The exception in a domestic environment is probably games consoles. It's not unusual for me to unexpectedly need a 50GB download and in that case speed is paramount. But for this I have it hard-wired directly into the router, that's not something I'd trust to Wi-Fi if I could possibly avoid it.
Honestly, does it matter?
No, not really - but it was sort of asked, and it's nice to know things.
The exception in a domestic environment is probably games consoles. It's not unusual for me to unexpectedly need a 50GB download and in that case speed is paramount. But for this I have it hard-wired directly into the router, that's not something I'd trust to Wi-Fi if I could possibly avoid it.
Even then, that download is presumably over the Internet, and so bottlenecked by the upstream. Only time that it makes a difference for me is if I'm doing network backups/restores.
No, not really - but it was sort of asked, and it's nice to know things.
I don't disagree, I'd want to know that things were working optimally. But I'm a geek and obsess about this sort of stuff, I think that for most 'normal' people if you can't tell the difference then it doesn't matter.
Even then, that download is presumably over the Internet, and so bottlenecked by the upstream. Only time that it makes a difference for me is if I'm doing network backups/restores.
Sure. But most people aren't doing bulk backups (or backups at all, sadly).
I'm fortunate enough to have symmetrical 500Mbps FTTP so for me the Wi-Fi is the bottleneck (which is why the Xbox is hard-wired and it absolutely canes downloads); up in the office I bought a high-gain USB Wi-Fi antenna and still 'only' get about half of that to the desktop. But I don't expect this to be typical - yet - for most people. My former boss lives in the middle of nowhere, on a good day he gets speeds akin to dial-up.
Or working. I do some 3D rendering and virtual testing. Ain't doing that over a slow connection.The exception in a domestic environment is probably games consoles.
Some of the error checking stuff for legal frameworks/systems gets a bit sniffy if network speed/ping gets slow or if you drop too many packets. I suspect they think your network is being interfered with.
Ironically enough, i have much faster connection at home with FTTP and WiFi than i do in the office with an ethernet connection...
Wired at home i get pretty much my full, paid for, network speed.
Or working. I do some 3D rendering and virtual testing. Ain't doing that over a slow connection.
I'm curious, how does that work? Have you got a local server or is it done across the Internet? The compute power requirements must be high but the image is 'just' an image, no? (Genuine question, I'm not being a nob, I'm interested.)
Have you got a local server or is it done across the Internet?Both, the models are rendered locally from server data, as it's a live database. But all the mechanics, rule based stuff and so on is handled on a central server. There is also an AI engine used to build the test cases, that runs on another server.
There is a "lite" version of the package, but it's got no AI engine and based on a weekly build. So ok for most, but i'm in the group developing it. So i need immediate and current access.
To be fair, i don't access it from home often, but when i do i need it to work.
I'm confused... It seems like all the heavy lifting is done by connecting to servers rather than physically running on the laptop?
In which case you just need a decent Internet connection.
Or are you connecting to a home server on your laptop via your local network?
Or are you downloading stuff to your laptop and then doing the compute on the laptop? A laptop is really the wrong tool for the job in that case. It would be like using a very good pair of scissors to try to cut down a tree...
Can you not just do it all via a remote connection to the server(s)?
Or are you downloading stuff to your laptop and then doing the compute on the laptop? A laptop is really the wrong tool for the job in that case. It would be like using a very good pair of scissors to try to cut down a tree...
Basically this, but the computer is also continually connected to make sure the physics and rules of the "game" aren't broken, and to make sure everything is synchronised properly.
It's a mobile workstation so more of a pair of scissors with a 30cc 2-stroke... But at least i can carry it.
Can you not just do it all via a remote connection to the server(s)?
Probably. But that's not the way it's been configured.
Have you got a local server or is it done across the Internet?Sorry, misread that, a lot of it is done across the internet from a server. My local server (i have one) is mine, and i'm not running work stuff on it!