Wireless faster than wired??

nst68 · 31 Dec 2020 at 13:24

I've run Cat5e cable from my router in to the office (wireless was rubbish there) and run some speed tests and i'm getting slower results than when I run the same test on my phone when it's next to the router

I'm on Virgin 350mb.

The cable run is
router --> switch --> switch --> switch --> laptop

The switches are all TP-Link gigabit, the max cable length is about 10m.

All the cables (apart from the router to switch) are ones I've made (I had to drill through two walls), I tested them with a £12 Amazon tester, but just this test that I've wired it up correctly and not how good the connections are?

I was expecting the wired connection to max out Virgin :confused:

shiver · 31 Dec 2020 at 13:31

Why so many switches daisy chained?

andy_mk3 · 31 Dec 2020 at 13:36

Have you tried plugging into different switches to see if there's a certain point at which the speed drops?

Terminal_Boy · 31 Dec 2020 at 13:37

Have you tried a different Speedtest app? Some open up a different number of IP sockets which gives different results.

That is an odd configuration...

ChrisD. · 31 Dec 2020 at 14:28

Try it directly from the VM hub or as close to as possible.

Glanza · 31 Dec 2020 at 14:36

Run a cable direct to the hub for testing, if you get full speed, your cables are either not 'proper' network cables or you've wired them up wrong when crimping.

nst68 · 31 Dec 2020 at 15:35

shiver said:
Why so many switches daisy chained?

1 by the router for smart switches, pi-hole and Hive hub
1 by TV (opposite side of the room from router) for TV and Xbox
1 in office (through wall from TV switch) for our work PC's.

Glanza said:
Run a cable direct to the hub for testing, if you get full speed, your cables are either not 'proper' network cables or you've wired them up wrong when crimping.

no idea what said:
Have you tried plugging into different switches to see if there's a certain point at which the speed drops?

I'll try direct to the router first, if that's full speed, then it's detective time.

Avalon · 3 Jan 2021 at 04:38

Generally you want to remove variables to identify the issue, however I would also look at different speed tests, for example the web clients do tend to vary, download the speedtest.net desktop app, this will allow you to log your test history and easily change test server (or more importantly keep the same one).

bledd · 3 Jan 2021 at 10:55

What NIC does the laptop have? Is it connected to power?

Could be 3rd party software slowing it down.

Run iperf3 on two machines between each other.

nst68 · 4 Jan 2021 at 10:34

bledd said:
What NIC does the laptop have? Is it connected to power?

Could be 3rd party software slowing it down.

Run iperf3 on two machines between each other.

It's a Realtek PCIe GBE Family Controller.

(yep it was connected to power when running the tests).

I tried a program called Lan Bench, with two laptops plugged directly in to ports on the router and that didn't produce consistent results.
I'll try iperf3

nst68 · 4 Jan 2021 at 11:24

can I use iperf3 between 2 directly connected pc's?

it looks like it uses IP addresses to define the computer to test with

ChrisD. · 4 Jan 2021 at 11:29

nst68 said:
can I use iperf3 between 2 directly connected pc's?

Yes, make sure the firewall allows the port it uses. Have you tried directly into the router?

nst68 · 4 Jan 2021 at 13:08

ChrisD. said:
Yes, make sure the firewall allows the port it uses. Have you tried directly into the router?

I used a program called Lan Bench (not iperf3) direct into the router for both laptops, but that gave significantly different speeds when I ran it about 5 times (using shop bought cables).

It's got to the stage where I'm not trusting any cables, so want to do be able to test cables direct between 2 laptops and then I can expand from there.

bledd · 4 Jan 2021 at 23:22

Here's how to use it.

https://iperf.fr/iperf-download.php

Get the 64bit latest Windows version

Extract it to a folder on your PC.

Go to this folder in Explorer, now click the address bar and type 'cmd' then hit enter.
It will open a CMD window in that folder.

Now type.

ifconfig (to find your IP address)

iperf3 -s

Now it's running in server mode

It should prompt Windows Firewall to allow it.

On your other PC.

Do the same..

But type

iperf3 -c IPofFIRSTpc

the-evaluator · 5 Jan 2021 at 07:38

bledd said:
On your other PC.

Do the same..

But type

iperf3 -c IPofFIRSTpc

iperf is generally more reliable if there's multiple streams running, so on the client I'd do 'iperf3 -c ip.of.ipserf.server -P 10 -t 60 -i 10' to run 10 parallel streams for 1 minute and to report the results every 10 seconds rather than every second.

Django x2 · 5 Jan 2021 at 10:37

Phone uses a compression algorithm too, so not a valid test. Wireless Laptop Vs same connected laptop is the only sensible test here.

And yikes to your switch setup. You might want to consider LACP or LA if you have enough ports on the switches or if they have SFP?

nst68 · 10 Jan 2021 at 16:08

the-evaluator said:
iperf is generally more reliable if there's multiple streams running, so on the client I'd do 'iperf3 -c ip.of.ipserf.server -P 10 -t 60 -i 10' to run 10 parallel streams for 1 minute and to report the results every 10 seconds rather than every second.

Ran the above command a few times and got these results

The server was wired to the router and the client was my work laptop in it's normal postition through the 3 switches.

There were 3 lower(ish) tests, the average with these was 793Mbps, without was 820Mbps.

Do these speeds look ok for cat5e? (they do to me).

Thanks

bremen1874 · 10 Jan 2021 at 17:08

With Cat5e and a single switch between machines, I get a consistent 949 Mbit/sec running the same test.

Avalon · 11 Jan 2021 at 04:24

Realtek would generally make me nervous here, but under Windows you should the best/least awful experience.

Armageus · 11 Jan 2021 at 07:14

nst68 said:
There were 3 lower(ish) tests, the average with these was 793Mbps, without was 820Mbps.

Do these speeds look ok for cat5e? (they do to me).

Yes depending on what else was using the switches uplinks at the same time