FTTP speed drop on LAN

Hi folks – hoping I might be able to get some help on a LAN issue at home. We had EE Full Fibre 300 installed last week which is running consistently at 250/45 at the router. I have a reasonably extensive home network – all cabled with Cat5e around the house and to some outbuildings (the garage and home office). In anticipation of the FTTP I upgraded all of the switches to gigabit capable and everything is mainly working fine in the house but not in the outbuildings. In particular to my office – this is a cable run of 100m (pretty much exactly) and is hard-wired through a gigabit switch into the back of my PC. The PC network adapter is saying connection is 1000/1000 so all good. Except the download speed test on the PC is reduced to about 90mbps (much lower than in the house at 250mbps) but the upload is the same as in the house at around 45mbps. My initial thought was speed loss over the distance (albeit I know 100m should have been OK) but that wouldn’t explain the upload speed which is not reduced, which it presumably would have been if this was the problem? I also have a wifi access point down in the office and this runs at the same speed as the PC so I can rule out it being a PC specific problem. 90mbps is till decent to work with me but it’s irritating me so any advice/thoughts very welcome!

Initial thoughts are:

1. On a full FTTP connection, the actual fibre speed should be much closer to 300 Mbps down and 50 Mbps up at directly connected at the router or connected via a wired gigabit capable switch.

2. The 100m Cat5e connection sounds problematic. If the cabling is 'good' there should be absolutely no drop in speed whether the link is 0.5m or 100m long.

Now by rights (although it is seldom to never done in a residential environment) - all fixed data cabling should really be performance tested to make sure it will support whatever services you want to run over it. Most people don't bother or their installers (themselves or their sparky) simply don't have accesss to the necessary test tools (like a Fluke DSX) to do so. But it really is the only way to verify if the cabling meets the actually Category performance limits.

Otherwise rather than testing from ‘the bottom up' you will need to try less ideal but alternate methods testing from ‘top down’ to try and determine if you have a cabling fault...

- Can you check and try with another port on the switch?
- If the switch is managed, can you log in and check the port stats; receive and transmit packet errors are never a good sign, also port 'flapping' or speed re-negotiations can indicate an underlying problem?
- If you place another switch, so a switch at both ends of the long link, does that improve matters?
- you could run a local speed test across your network to check. Various tools exist - I've used TamoSoft Throughput Test and LAN Speed Test from TotuSoft in the past successfully.

Edited by Pheasant (Wed 07-Apr-21 19:59:35)

Many thanks - really helpful. To answer your questions:

1). I will try other ports on the switch and see if that makes a difference
2) Switches are unmanaged Tp-link - at the basic end but speeds in the house go through switches also so don't think they will be the problem. I do have a spare managed switch at work that I could 'borrow' but think it is only 10/100 so not much help!
3) The long link does have a switch at both ends. The router goes into a 16 port switch at the house end with patch panel connections going to various destinations. The office actually goes to another with/patch panel in the attic and then onto the office. At the office end it arrives into a switch which is then patched to a couple of RJ45 sockets (one for wifi access point, one for my PC, one for a TV) so when I think about it there are 3 switches on its journey - perhaps that might be part of the problem. I should be able to test that by patching the office straight from a LAN port on the router and avoiding all of those switches - will be very interesting to see what difference that makes.
4) I've not used LAN Speedtest tools so thanks for those recommendations

I guess if none of the above makes a difference then the issue likely to be the cabling - self-installed and a hell of a job last summer! It's SWA armoured Cat5e buried about 500mm down so if that does end up being the problem I suspect I might have to live with it as I won't be popular ripping up the garden again!

Thanks that clarifies a few things. If your connected as follows:

ONT——Router——Switch——-/100m link/——-Switch——PC/access point

Then if the last switch is 1000M the PC or access point will just see a 1000Mbps link.

On the other hand:

ONT——Router——Switch——-/100m link/——-PC/access point

You should see what port speed the PC negotiated with the switch at the house. It’s a more honest test of the 100m Cat5e link.

Do you have a basic, cheap tester? If not, one similar to https://www.ebay.co.uk/itm/Professional-RJ45-RJ11-Ne... will do the basics and worth buying for £5-6 - there are plenty of sellers.

It could be that you have one conductor that is badly terminated or broken - that would be enough to restrict it to 100Mbps.

Thanks - I do have a tester - the last time I checked (admittedly not recently) it was testing 1 to 8 fine. I will check again though to ensure nothing has changed.

And have you used teh correct pin assignment?

Just thinking out loud on various things....

End to End Channel vs Links. By the OPs description above there’s some patching going on, so it’s important to check each fixed link individually and then test together as a ‘channel’ with any patch leads in play. Faulty patch leads are surprisingly common. Swapping leads may uncover the problem.

Length. Going beyond 100m channel length (when patch leads etc. included) also isn’t advisable, although it should still work and I know people that overdrive links well beyond 100m. The cabling standards are all based on 90 metres of solid core cable in the permanent link plus up to a total of 10 metres of stranded core cable in the patch leads.

Performance. There’s also no guarantee that continuity / wire-map alone is the issue. But it is fundamental. Cabling performance limits come into play at longer cable lengths, especially pushing at 1000BaseT. With the OPs config, the switch to switch link could be dropping back to 100BaseT if there are problems on the link. The OP would be unaware as he’s connecting to the switches at either end so just seeing 1000BaseT back to the NIC in the pc.

Totally agree.

My comments are based around the very basic testing/checks that can be undertaken and will frequently show up an issue

Thanks both. I was watching TV and thinking about all this again so decided to get up and do some testing. I managed to patch a direct route straight through from the router to the office pc with no switches (and annoying the kids in the process!) and hey presto 310mbps in the office! So cabling fine. Then started adding switches back in and there is one that is causing the problem. It’s gigabit POE as has some cctv cameras attached but it is limiting traffic down to 100mbps. So I will get a replacement and should be good to go. Many thanks for the input - learnt a lot.