Reliable speed test for FTTP (900/100)?

Hi folks,

We have had FTTP installed for a couple of days now, using the BT Whole Speedtest site I regularly see over 1000Mbps down and around 140Mbps up but when I check other sites I only see around 250Mbps / 100Mbps is this something I am doing wrong?
I was expecting my Router to be a bottle neck (AirPort Extreme AC) but with the BT site showing those speeds can that be the case?
The PC I am testing on is using Ethernet and the engineer said Cat5e cables would be sufficient (I have some unopened Cat7 labeled cables to use if required).

Cat5e should be fine - unless you are 50m away.

The basic "no-frills" test at fast.com is normally relaible and the TBB one too which graphs performance with links to share that will allow others to comment.

Thanks, here is the result from TBB

That looks as though something is maxing out.


If you are using Windows, run TASK MANAGER and whilst not precsie it will give a good insight. Try it with BT Wholesale, TBB and fast.com see what it gives for all three.

BT Wholesale: 1008.73 / 146.92
fast.com: 260 / 110


Is the BT Wholesale site telling me lies?

Edited by D_an_W (Fri 19-Nov-21 18:38:52)

Graphical, browser-based speedtests can tax your PC, and the BT one may be less taxing. Hence the suggestion to look at your CPU usage: if it flatlines at 100% then your PC is the bottleneck.

Often Linux performs much better than Windows on the same hardware. You could download an Ubuntu Live USB stick, boot from that, and see if you get better results. Then know where to start looking - maybe Windows NIC drivers, maybe antivirus.

This unfortunately is the problem with gigabit services: they highlight problems with your client devices and your local network. (I have 300M).

In reply to a post by D_an_W:

I was expecting my Router to be a bottle neck (AirPort Extreme AC) but with the BT site showing those speeds can that be the case?
The PC I am testing on is using Ethernet and the engineer said Cat5e cables would be sufficient (I have some unopened Cat7 labeled cables to use if required).

Tell us a little bit about the PC you are testing with; OS/version, make/model, age and how it’s connected to the network, i.e. built-in NIC or external USB adapter (can sometimes be problematic too)

Thanks folks,

The PC is relatively old now but still fine for my needs (Until now maybe!)

This is the info I use for a Signature on another forum:

* CPU: Intel i7-3770K @ 4.2GHz * Cooling: Corsair H80i v2 / Noctua NF-F12 x2 * MoBo: Gigabyte GA-Z77X-UD5H (Rev. 1.1 - BIOS F16g Mod06) * RAM: Crucial 1600MHz DDR3 32GB (4x8GB) *
* PSU: Seasonic 760W Platinum * GPU: Intel HD 4000 (768MB) * Mon 1: Dell U2410 (A07) * Mon 2: Panasonic TX-P55VT50B * Cam: Logitech C910 HD * BD: Pioneer BDR-208DBK *
* HD 1: Western Digital WDC WDS100T2B0A 1TB * HD 2: Western Digital WD Black 10TB * WiFi & BT: Apple Broadcom BCM94360CD (802.11 a/b/g/n/ac - Bluetooth 4.0 *

Connected via one of the two built in Ethernet ports (Using the Atheros chipset).

It is actually running macOS Big Sur 11.6.1 which also happens to be the last this hardware will support without patching etc.

I have been keeping an eye on price of the TP-link Archer AX6000 as it looks like it might suite my needs but it's the BT Wholesale site showing me those speeds that makes me wonder if I need to replace the AirPort Extreme at all.

Edited by D_an_W (Fri 19-Nov-21 19:35:54)

In reply to a post by D_an_W:

makes me wonder if I need to replace the AirPort Extreme at all.

It is the first component I would swap out to test. They are quite old now. Is this the one?
https://support.apple.com/kb/sp680?locale=en_GB
That is the 6th generation and it went on sale in 2013, and Apple discontinued the whole lineup in 2018. If you have another router (e.g. ISP box) give it a try for a test.

My guess is the CPU in the Extreme is not up to the needs of PPPoE over your fast FTTP service, and Apple designed their routers for the USA market which don't typically use PPPoE.

Edited by jchamier (Fri 19-Nov-21 20:02:08)

In reply to a post by D_an_W:

It is actually running macOS Big Sur 11.6.1 which also happens to be the last this hardware will support without patching etc.

It's a Hackintosh? Definitely try booting into Linux as a comparison.

But once you've convinced yourself that it's not the ISP that's the issue but your own local machine, you can forget about it. Like you said, it's fast enough for what you need. It can't consume a gigabit reliably on a speedtest? Speedtests aren't what you actually use the machine for.

Are you able to connect the PC directly to the ONT? You will need to set up a PPPoE connection under Network with the login details.

That will at least eliminate the Airport from the equation and give you a baseline of what the hardware is capable of.

In reply to a post by jchamier:

In reply to a post by D_an_W:

makes me wonder if I need to replace the AirPort Extreme at all.

It is the first component I would swap out to test. They are quite old now. Is this the one?
https://support.apple.com/kb/sp680?locale=en_GB
That is the 6th generation and it went on sale in 2013, and Apple discontinued the whole lineup in 2018. If you have another router (e.g. ISP box) give it a try for a test.

My guess is the CPU in the Extreme is not up to the needs of PPPoE over your fast FTTP service, and Apple designed their routers for the USA market which don't typically use PPPoE.

Thats the one yes, can't complain at all for the length of time I have had it. Still unsure about what to replace it with but I think ASUS are ruled out as they only seem support 32 DHCP reservations.

In reply to a post by Pheasant:

Are you able to connect the PC directly to the ONT? You will need to set up a PPPoE connection under Network with the login details.

That will at least eliminate the Airport from the equation and give you a baseline of what the hardware is capable of.

I did try that after reading a post on here but couldn't get it to connect, maybe I did something wrong.

In reply to a post by candlerb:

In reply to a post by D_an_W:

It is actually running macOS Big Sur 11.6.1 which also happens to be the last this hardware will support without patching etc.

It's a Hackintosh? Definitely try booting into Linux as a comparison.

But once you've convinced yourself that it's not the ISP that's the issue but your own local machine, you can forget about it. Like you said, it's fast enough for what you need. It can't consume a gigabit reliably on a speedtest? Speedtests aren't what you actually use the machine for.

Hackintosh yes, been running macOS on it for quite some time now (Still have VMs with Windows of course!) I think I can remember how to create a Ubuntu USB boot drive, will give it a go thanks.

In reply to a post by D_an_W:

Thats the one yes, can't complain at all for the length of time I have had it. Still unsure about what to replace it with but I think ASUS are ruled out as they only seem support 32 DHCP reservations.

Possibly an old limit.
I've had over 32 set on my RT-AX86U. It says the max is 64 in the GUI.

Edited by j0hn83 (Fri 19-Nov-21 21:17:22)

In reply to a post by j0hn83:

In reply to a post by D_an_W:

Thats the one yes, can't complain at all for the length of time I have had it. Still unsure about what to replace it with but I think ASUS are ruled out as they only seem support 32 DHCP reservations.

Possibly an old limit.
I have over 32 set on my RT-AX86U. It says the max is 64 in the GUI.

Interesting thanks, it's only information I read in the manuals so could well be outdated.

In reply to a post by D_an_W:

In reply to a post by Pheasant:

Are you able to connect the PC directly to the ONT? You will need to set up a PPPoE connection under Network with the login details.

That will at least eliminate the Airport from the equation and give you a baseline of what the hardware is capable of.

I did try that after reading a post on here but couldn't get it to connect, maybe I did something wrong.

Worth doing. Not particularly difficult in MacOS:
- click '+' to create a new service
- choose 'PPPoE' (second from bottom)
- select the wired ethernet port from the drop down (it defaults to this anyway)
- Give it a name and click 'create'

Once its built, select the new PPPoE service
- enter your account name + password (tick remember the password)

Should be ready to connect from that point, using the button

This gives you direct unfettered access at the ONT, so no routers etc to get in the way. Should give you the cleanest test result.

I really wouldn't expect much difference performance wise doing it this way in MacOS and Ubuntu, certainly not as much as "abstraction" as doing it with Windows.

By the way who is your ISP? Did you test with their supplied router rather than your AirPort in place?

Have you tried speed tests with any other devices besides your hackintosh?

What about doing a simple (non-browser) speed test using the ookla speedtest.net app for Mac?

In reply to a post by Pheasant:

By the way who is your ISP? Did you test with their supplied router rather than your AirPort in place?

Have you tried speed tests with any other devices besides your hackintosh?

What about doing a simple (non-browser) speed test using the ookla speedtest.net app for Mac?

My ISP is Aquiss who don't supply a Router, don't have any other Ethernet devices I can test with at the moment and yes I did try the Speedtest.net app and got a similar result to their website.

Just trying the PPPoE connection again.

It's the AirPort Extreme that's slowing you down. I have 900/110 FTTP as well. Using the BT Smart Hub2 I get full speed on a wired 2017 iMac. I changed the router to TP AX10 and my speeds were the same but the router was reliable and wifi way better.

On the AirPort Extreme acting as the router, I topped out at about 300Mb.

I still use the AirPort Extreme in bridge mode to extend my network via ethernet. That works fine and get full speed to any device wired to it.

I hope that helps.

Shrey

In reply to a post by Pheasant:

In reply to a post by D_an_W:

In reply to a post by Pheasant:

Are you able to connect the PC directly to the ONT? You will need to set up a PPPoE connection under Network with the login details.

That will at least eliminate the Airport from the equation and give you a baseline of what the hardware is capable of.

I did try that after reading a post on here but couldn't get it to connect, maybe I did something wrong.

Worth doing. Not particularly difficult in MacOS:
- click '+' to create a new service
- choose 'PPPoE' (second from bottom)
- select the wired ethernet port from the drop down (it defaults to this anyway)
- Give it a name and click 'create'

Once its built, select the new PPPoE service
- enter your account name + password (tick remember the password)

Should be ready to connect from that point, using the button

This gives you direct unfettered access at the ONT, so no routers etc to get in the way. Should give you the cleanest test result.

I really wouldn't expect much difference performance wise doing it this way in MacOS and Ubuntu, certainly not as much as "abstraction" as doing it with Windows.

BINGO!



fast.com : 1.0Gbps / 99Mbps
speedtest.net: 909Mbps / 98Mbps

The problem was I had to manually assign an IP to allow the PPPoE to connect, once it did things look much different which as I used the same Ethernet lead that was connected to the Router puts the Router as the culprit as suspected?

In reply to a post by shrey:

It's the AirPort Extreme that's slowing you down. I have 900/110 FTTP as well. Using the BT Smart Hub2 I get full speed on a wired 2017 iMac. I changed the router to TP AX10 and my speeds were the same but the router was reliable and wifi way better.

On the AirPort Extreme acting as the router, I topped out at about 300Mb.

I still use the AirPort Extreme in bridge mode to extend my network via ethernet. That works fine and get full speed to any device wired to it.

I hope that helps.

Shrey

Thanks, just discovered that as you posted!

Yup you have found your culprit (and the post above yours also points the finger at the AirPort)....

Anyhow it's always good (in my mind anyway) to prove this beyond a doubt on your own setup. You could have been more unlucky and had further issues with the hackintosh, but that looks fine

Edited by Pheasant (Fri 19-Nov-21 22:19:05)

In reply to a post by Pheasant:

Yup you have found your culprit (and the post above yours also points the finger at the AirPort)....

Now to decide wether or not to pull the trigger on this TP-Link Archer AX6000

In reply to a post by candlerb:

This unfortunately is the problem with gigabit services: they highlight problems with your client devices and your local network. (I have 300M).

This needs to be a strap-line or something It is unfortunately going to become more and more commonplace

Cate5e is good for ~300ft or 100m at 1Gbps. In fact it is good for 2.5Gbps at that distance.

Should be ... Have recently come across some where either due to bad install or cheap cable it was not always possible at well under 100m.

In reply to a post by D_an_W:

(I have some unopened Cat7 labeled cables to use if required).

One born every minute. So which IEEE 802.3 standard do you think requires the use of Cat7 cable? That would none what so ever. I also very much doubt you have the proper connectors on the end of the Cat7 cable and you absolutely do not have any equipment that could properly terminate those connectors, because none has ever been released. In fact it worse than that because if you don't have the proper connectors on the end of the Cat7/8 cable you actually make things worse not better. If you don't understand why then it is pointless trying to explain to you why.

Anything beyond Cat6a is for suckers and those who like to engage in willy waving contests while burning the planet.

Cat5e is a 20+ year old, mature (some would say outdated) cabling standard with relatively, modest and yet forgiving performance limits (100 MHz) you can get away with quite a lot… Yet there always remains a world of difference between say…

a) Dedicated installer, reputable/warrantied single ‘cabling system’ supply from one of the wel known structured cabling manufacturers. Every perm. link performance tested with a certification tester like a Fluke DSX. Hundreds or thousands or points.

b) Part timer/sparky, cabling of unknown provenance, mismatched everything (whatever the wholesaler had behind the counter) and installer that doesn’t really know what they’re doing, Testing?? what’s that I just bell’d out the pairs guv…a few dozen maybe a hundred points. Cowboy territory.

Driving cabling at 20 or 30m is pretty easy. Performance of installed cables at the extremes of 90m sorts the wheat from the chaff (especially beyond Cat5e), but you never see those lengths outside of large commercial installs anyway.

In reply to a post by jabuzzard:

In reply to a post by D_an_W:

(I have some unopened Cat7 labeled cables to use if required).

One born every minute. So which IEEE 802.3 standard do you think requires the use of Cat7 cable? That would none what so ever. I also very much doubt you have the proper connectors on the end of the Cat7 cable and you absolutely do not have any equipment that could properly terminate those connectors, because none has ever been released. In fact it worse than that because if you don't have the proper connectors on the end of the Cat7/8 cable you actually make things worse not better. If you don't understand why then it is pointless trying to explain to you why.

Anything beyond Cat6a is for suckers and those who like to engage in willy waving contests while burning the planet.

Wow,

Did I catch you at a bad time or something?

I wrote "Cat7 labeled" and left them unopened as I was sceptical about them myself but thanks for making it clear and no, don't bother explaining anything as I doubt I would read anything you write.

In reply to a post by Pheasant:

In reply to a post by candlerb:

This unfortunately is the problem with gigabit services: they highlight problems with your client devices and your local network. (I have 300M).

This needs to be a strap-line or something It is unfortunately going to become more and more commonplace

Thanks for yours and others help with this, much appreciated.

No worries. Good luck with the new router…and enjoy the speed!

In reply to a post by Pheasant:

No worries. Good luck with the new router…and enjoy the speed!

Thanks, the price just dropped too so placed an order for one.

In reply to a post by D_an_W:

Interesting thanks, it's only information I read in the manuals so could well be outdated.

The US site Small Net Builder has a pretty good technical forum on Asus routers, and the third party firmware from Merlin. (I'm running an AX88U).

Try a test at speedtest.net
It should detect you're on a gigabit connection and suggest you download their desktop app for Windows... It gives far more consistent results, than an in-browser test.
For the server; I normally choose University of Oxford IT services and my average speedtest (using that app and the Oxford server) is something like 910Mbps/109Mbps

In reply to a post by adebov:

Try a test at speedtest.net
It should detect you're on a gigabit connection and suggest you download their desktop app for Windows... It gives far more consistent results, than an in-browser test.
For the server; I normally choose University of Oxford IT services and my average speedtest (using that app and the Oxford server) is something like 910Mbps/109Mbps

Thanks for the tip, I have a new Router arriving tomorrow so will retest once setup.

In reply to a post by adebov:

Try a test at speedtest.net
It should detect you're on a gigabit connection and suggest you download their desktop app for Windows... It gives far more consistent results, than an in-browser test.
For the server; I normally choose University of Oxford IT services and my average speedtest (using that app and the Oxford server) is something like 910Mbps/109Mbps

I use their CLI App and run it from console and get my full speed.

Paul

With the new Router installed I now get around 917/110 when running a test.

Great!👌👍

In reply to a post by D_an_W:

With the new Router installed I now get around 917/110 when running a test.

Nice

Paul

In reply to a post by Pheasant:

Driving cabling at 20 or 30m is pretty easy. Performance of installed cables at the extremes of 90m sorts the wheat from the chaff (especially beyond Cat5e), but you never see those lengths outside of large commercial installs anyway.

I think links closer to the limit are a bit more common now in domestic settings with the rise of the "garden office". That said it is probably a single link and much easier to get one port working at distance, because if it is not working it will be immediately apparent.

Oh and I would note a lot of managed switches and for that matter better quality ethernet adaptors can do comprehensive cable testing if you know where to look to enable it. I remain convinced there is a significant market for a LinkRunner type device at a fraction of the price of a LinkRunner by exploiting the cable testing features of a suitable ethernet chipset. I am lucky to have a LinkRunner Pro that I salvaged from being thrown out at work, cleaned up the mess from the leaky batteries, replaced a battery contact that had been corroded away and fixed a dodgy connection on the RJ45 socket.

What sort of in-switch testing have you seen/experienced? I’ve only seen very basic (as you’d expect) TDR tests that can measure cable length, and open/short pairs. That’s it. I wouldn’t really say that’s comprehensive - certainly not when compared to an objective, dedicated cable test tool. On the other hand against a basic continuity tester they are OK.

Fair point on external cable runs, but personally I have a pathological hate of running structured cabling between buildings externally. It’s done, I’ve done it took, but it’s [censored] really. Much safer and better to run a short pre-term armoured fibre (like the GI Invacom use for fibre-IRS TV) and pop the ends into a baby switch or media converters. But we’re getting way of course…