Sudden slow wifi with fibre broadband

I have a puzzling problem here with sudden slow wifi that I'm struggling to solve and wondered if anybody might know the answer. For the last 3-4 years I had excellent speeds anywhere in the house that were pretty much hitting the maximum. The router is downstairs next to the master socket, but upstairs in the office I would typically see wifi download speeds of 70+mbps day and night on a 76mbps line. Switching to Ethernet made no real difference.

A couple weeks ago I changed ISP, but on the first day there was an issue with the router account login that connects to the ISP, so I was provided with a temporary one and it all worked great. Download speeds were similar to before and ping times were a bit faster.

After a couple days some changes were made at their end and there was 10 minutes of downtime before the internet came back. When I checked the router I could see they had removed the temporary login and added the correct one. From that moment on I've had pretty useless and erratic wifi download speeds, though upload speeds are generally hitting the maximum and rarely go less than 18mbps.

In practice if I plugged in the Ethernet cable speeds were fairly close to the router sync speed, but the moment I tried wifi it would drop by at least 30mbps, even from a couple meters away. By the time I went upstairs it would be all over the place, but generally slow and at times it's been under 5mbps. I swear nothing changed in my house during that 10 minute switchover period...

I tried a couple different routers with no improvement and the ISP sent me a new router with a 5Ghz band to see if that helped. At a certain point it all became very unstable as the router sync speed progressively slowed, while I was losing the DSL connection, along with the phone line repeatedly. The router sync speed had dropped to around 47mbps. It's possible that changing routers a couple times was messing with DLM, but I'm surprised if that was also affecting the phone line.

Yesterday somebody from OpenReach visited and checked out the line. He reset DLM and verified that the line reaching the house was all good. It measured 78mbps at the socket. He did exchange the plate for a newer version, saying it would produce less noise on the line.

The router started to sync at 71mbps, but with a 1m Ethernet cable that already dropped another 6mbps. Once we switched to wifi in the same room the speed dropped back to around 30mbps like before. I checked upstairs with the engineer on another computer and it was down to 15mbps. This is a modern 3 bedroom house, so we are not talking huge distances here and the wifi signal is always on maximum, whether on 2.4Ghz or 5Ghz. In testing 5Ghz generally performs better than 2.4Ghz.

I used to get 15mbps on wifi in this room back in the days when I had ADSL and now I'm on a business fibre package with a fixed IP that should be rock solid. The router is an Asus DSL-N66U that should be pretty good and certainly no worse than routers I've used in the past.

I've contacted Asus to ask if they have any recommendations for the wifi settings, but I am really at a loss to understand this. The wifi channels change automatically to avoid congestion and I can see with a wifi analyser that there are no other routers on the same space. I've also tried using Mac filters in case the neighbour's were helping themselves to our bandwidth, but this is not a congested area and they are out all day.

The DSL connection appears to have remained stable since the OpenReach visit yesterday, so I am reluctant to switch anything off, but I'm wondering if something can be improved in the wifi settings. I've tried moving the router and furniture previously to no avail, but I keep coming back to the point that everything was great on wifi for the last few years until this switch by the ISP that occurred a couple days after joining. They tell me this cannot be the cause of the problem.

Can anybody think what may be going on here?

You need to do some testing of speed between devices on the network if trying to troubleshoot wifi issues, since with the variations you've had in connection speed you will have been getting a confusing picture, plus variations in how the ISP is performing maybe.

When you say the Wi-Fi is on maximum all the time what do you mean?

By maximum I just meant I was seeing the best possible speeds that were pretty much inline with the 76/19 limits for this line. It was quite normal to see 73mbps downloads on wireless for the last few years. I have full strength signals everywhere and fibre broadband has always performed really well here until a couple weeks ago with this change.

I've tested this now on three different Windows laptops, plus a Mac Pro, Android smartphones and a Kindle HD tablet. All of them are showing similar results with this sudden drop in wifi speeds. Is there some specific test I could try that might help to identify what is happening here?

I ran a test just now at http://internethealthtest.org and my average download speed is 11mbps, while the upload speed is over 18mbps.

And that is just another speed test, as we wrote http://www.thinkbroadband.com/speedtest we have made sure it stores as much info on the test as possible so the graphs are important.

If you aren't going to set up internal file transfers for testing wireless, then you need to compare wireless testing with the same devices over Ethernet to the same speed test site

It could be that your wireless is still perfect, but the performance on the ISP is more variable than a previous one

OK I just registered on the main site and ran a wifi test with the link you provided. The results were not hugely different to the other speed tests I have tried.

I am quite happy to run internal file tests for transferring files. It's just not something I've needed to test before and I'm not sure how best to do this to produce useful results.

I know from yesterday's visit by OpenReach that 78mbps is arriving at the main socket and I can see from the router control panel that it's currently synching at 71mbps. If my new ISP is somehow slowing it down to this degree that would be atrocious.

You have not mentioned the ISP.

If you have switched to Sky, yes their router is known to be poor on wireless, even the Q router isn't the best since it forces you to use channel 36 on wireless with 5ghz.

No it's not Sky. The router I've been supplied with is an Asus DSL-N66U, which should be pretty good in theory. I've noticed that it changes channels, so evidently that part is working.

I've just run the 1 gig download test from this website and it took just under 12 minutes. That's really slow for a 76mbps fibre connection.

Edited by charlestown (Sat 18-Mar-17 16:04:24)

What Wifi channels are the old ISP and new ISP routers using, and are you sure that you don't have any conflicts from nearby properties?

Acrylic Wifi for Windows will show all Wifi networks in range. There are similar applications for Android.

I think both the old and new router automatically choose channels based on obtaining the best performance. The old one was from BT, so I don't know if they have any kind of restrictions. The Asus is currently on channel 40.

I just checked using a wifi analyser app on the Kindle HD tablet and I could see a channel overlap with one other router nearby on 2.4Ghz, but on 5Ghz mine was the only one visible with empty channels on both sides. It looks like it is broadcasting at 2.4Ghz and 5Ghz simultaneously, possibly for the sake of different devices, but that was never a problem previously.

As things stand the wifi performance was better with the old ISP than I am seeing with the new one on Ethernet. In theory that could improve, but there is no sign of that as yet.

Just an afterthought but we have two DECT phones. The main one with the base unit is connected to a phone socket in an upstairs bedroom, while there is a second charger point for the other phone in the same room as the router. This has been like that for years, so I would be surprised if it was suddenly causing problems.

Have you tried naming the 5ghz and 2.4ghz SSIDs different? Then only connect into the 5ghz name

Yes they are named differently. On this computer for example I'm connected to the 5Ghz signal, but my phone and one of the laptops will only connect to the 2.4Ghz. This was never a problem previously.

Edited by charlestown (Sat 18-Mar-17 20:53:09)

In reply to a post by MrSaffron:

so the graphs are important.

^^^ This bit

In reply to a post by charlestown:

Yes they are named differently. On this computer for example I'm connected to the 5Ghz signal, but my phone and one of the laptops will only connect to the 2.4Ghz. This was never a problem previously.

What router did you have before? Can you not go back to it?

For the last few years I was using a BT HomeHub 5, but that only works with BT. I also had a BT modem with a Tp-Link router that was fine when I was with BT.

The new ISP originally gave me another Asus that was limited to 2.4Ghz. This had the same problem, though it was perfect for the first two days while I had that temporary login, before they made that switch I mentioned in the original post. Everything changed in 10 minutes when that change occurred.

I am really doubtful that the router is the problem, but it may well be the wifi settings for their setup. Perhaps there is some obscure setting that is causing problems.

Edited by charlestown (Sun 19-Mar-17 09:04:46)

Can you run a thinkbroadband speedtest and post the full link to it?
e.g here is one of mine
http://www.thinkbroadband.com/speedtest/results.html...

We need to see the full page and graph.

Sure here is one: http://www.thinkbroadband.com/speedtest/results.html...

As you've not linked to the result I could but won't in case there is some reason you are not wanting to name the provider but would say

The shape of the lines suggest there is a problem, and if you have a similar shape graph when using an Ethernet connection then someone is wrong and it could well be the capacity that the provider has.

Noticed you were running the flash test, so trying http://labs.thinkbroadband.com/speedtest/ as a double check (i.e. does not use flash) is worth while.

This example http://tbb.st/1489941313624952455 shows a very good test, notice how the two download tests are almost identical, and after a very short time the is running flat out.

I did post a link, but it was elsewhere in the thread. I think it's clear I'm not simply placing blame on the ISP but trying to find a solution. I work from home and rely on having a fast, stable connection.

This is wifi from yesterday with the flash test using Chrome:

http://www.thinkbroadband.com/speedtest/results.html...


This is the same test now using Chrome, but on Ethernet and this is pretty slow I think at 49mbps on a 76mbps connection.

http://www.thinkbroadband.com/speedtest/results.html...


This is the non flash test using an Ethernet cable on Chrome

http://www.thinkbroadband.com/speedtest/results.html...


This is the non flash test on wifi at 5Ghz on Chrome.

http://www.thinkbroadband.com/speedtest/results.html...


These are actually faster than I generally see when testing with speedtest.net, but nevertheless my Ethernet connection here is slower than the wifi connection I had on the old ISP. Hopefully the graphs will tell us something useful.

Who is your ISP?

It's CloudScape Connect.

Can you run 2 speedtests, all over ethernet please.

Also run 2 speedtests on www.speedtest.net please - again ethernet.

Honestly those graphs are really bad and they indicate very strongly congestion (looking at the wired one). I can go into more details, but effectively it means the ISP is not purchasing enough capacity for everyone as they're keeping costs down. You rarely see this on big ISPs (BT, talktalk, sky) nowadays but it used to be a lot more common.

How long is the contract for?

Edited by ukhardy07 (Sun 19-Mar-17 17:58:22)

OK here are two from speedtest.net on Ethernet:

http://www.speedtest.net/my-result/6144702183

http://www.speedtest.net/my-result/6144705297

And two from this site:

http://www.thinkbroadband.com/speedtest/results.html...

http://www.thinkbroadband.com/speedtest/results.html...

I am actually on an unlimited business package that is more expensive than the home packages and supposedly free of congestion. It comes with a fixed IP address. The contract is for 12 months and the internet was fast for the first two days while I had a temporary login. The moment they switched that it has been pretty slow and unreliable.

All the troubleshooting has been very time consuming. There have been times with the wifi when I've seen download speeds below 5mbps.

It shows in the link he posted in reply to your example.

Odd that two people don't seem to see it.

I suggest you check with them that they haven't accidentally dropped you onto a standard non-business connection when they changed whatever it was. Before chasing what may not be a fault so passing all tests.

That seems very reasonable and may explain the problem, so I'll ask them now. For comparison sake this is a speed test I once ran with BT.

https://sunnymede.net/screenshots/2453051347.png

I've just had a reply from the ISP and they think the results here are not reliable. They ran a check on a Virgin cable connection that showed good results at speedtest.net but much worse on the checker at this site.

http://www.thinkbroadband.com/speedtest/button/14900...

http://www.speedtest.net/my-result/6146039174

Speedtest.net (OOKLA engine) is well known to be ISP's preferred tester. It drops the highest and lowest few readings and averages the rest, always multi-threaded.

Try this one. Ignore all the red instructions, just say you've done them. Of course make sure nothing else is using the connection. Further Diagnostics won't work but you should get speeds and latency.

Thanks I'll give this a go, since I don't seem to be getting very far with Cloudscape. I have lost so much time on this over the last couple of weeks...

Edited by charlestown (Mon 20-Mar-17 12:03:41)

We have been in constant contact with this customer, replaced his router with a more expensive (ASUS DSL-N66U with 5Ghz WiFi) device at our expense.

The main issue is his WiFi speed it slower than his wired speed and this drops off as his distance from the router increases. Now as far as our understanding of WiFi does this is normal.

If anyone would like to add to this thread and has anything to add please contribute.

Cloudscape Connect Ltd

So ignoring speed tests, how is the connection feeling?

That Ookla test was to a Newbury server, our testing is in London and will be consistent, whereas Ookla will pick the closest in terms of latency and thus more likely to give a higher result, but may change from day to day.

The usual defence of providers is our tester is WRONG, when the reality is that it is designed to reflect user experience, and not just report what is the connection speed.

Also no reason why people on Virgin Media should not get decent speeds an example grabbed from recent tests
http://tbb.st/1488378290435211555

And we can go a lot faster too http://tbb.st/1489229024125180055

So does the ISP consider all four Ethernet based tests in the post at http://forums.thinkbroadband.com/technical/t/4536103... to be perfectly good for the service purchased?

Some examples from other Cloudscape customers

http://tbb.st/1489361704190764755
http://tbb.st/1488576076452781855
http://tbb.st/1489870501497483755
http://tbb.st/148986757068348282744
http://tbb.st/1489923014638151155
http://tbb.st/1489862855693761355
http://tbb.st/1489858376199175555
http://tbb.st/1489857708388962655
http://tbb.st/1489855836891899855

Hi Andrew,

We would expect around 65+ M download speed on his cct, he is in Sync at 70 M. When we have had a remote desktop session with his that is what we were seeing.

The main dispute is WiFi speed, we believe that on average you lose between 20% and 40% throughput on WiFi on 2.4Ghz, less on 5Ghz, and this loss increases with distance. On 5Ghz the distance/speed loss is greater and environment etc can effect this.

We have sugested that the customer uses a Power-Line adaptor to reach his laptop which is normally situated in another room to the router.

A very simplistic view, its not a blanket 20 to 40% drop off on 2.4GHz.

From what we can see on the speed test results over Ethernet there is an underlying issue beyond just Wi-Fi

Hi Andrew,

We have had a SFI visit, the NTE was upgraded and his profile reset. He is in SYNC at 70M, and he is on a new under-contended interconnect on our network.

We can refer this back to the carrier to check throughput on his Port but this does not resolve his gripe about WiFi performance. When we have had a remote desktop session with a wired connection and WiFi disabled we could not identify any issues.

The customer is adamant that he was getting 70M on WiFi through out his premises and now it is much less. I have no explanation.

Here is a test from another cct on the same interconnect in SYNC at 33M / 7.5M
http://www.thinkbroadband.com/speedtest/button/14900...

And is that test a good one?

In reply to a post by cloudscape:

Hi Andrew,

We would expect around 65+ M download speed on his cct, he is in Sync at 70 M. When we have had a remote desktop session with his that is what we were seeing.

The main dispute is WiFi speed, we believe that on average you lose between 20% and 40% throughput on WiFi on 2.4Ghz, less on 5Ghz, and this loss increases with distance. On 5Ghz the distance/speed loss is greater and environment etc can effect this.

We have sugested that the customer uses a Power-Line adaptor to reach his laptop which is normally situated in another room to the router.

Re this reply to Andrew, you were asked about the wired results from four tests yesterday, from four testers.

See two tests I have just conducted, wired, for comparison.

Thinkbroadband
and BT Wholesale Performance Test:

Download speed achieved during the test was - 61.22 Mbps
For your connection, the acceptable range of speeds is 40 Mbps-63.44 Mbps .
Additional Information: IP Profile for your line is - 63.44 Mbps

Upload speed achieved during the test was - 11.93Mbps
Additional Information: Upstream Rate IP profile on your line is - 20 Mbps

Mine is a Home product, not a business one, and the tbb graph is how any properly performing decent quality xDSLx should look. On a Home/SOHO one even Plusnet can achieve similar, though not consistently.

Now a wireless tbb from downstairs near a side wall, with the router upstairs near the opposite side wall. The house is built completely metal-framed, then brick-clad. So joists are quite a barrier.

Sync speeds in my sig.

The point of the thinkbroadband test is that multi-threaded it is easier to achieve high throughput. Single-threaded being around the multi-threaded line indicates adequate capacity at the ISP end. A large gap and wildly fluctuating graph lines suggest there may be a capacity problem at the time of the test.

That's why I suggested to the OP to check he hadn't accidentally been switched to a Home product, where on a smaller ISP that could perhaps be expected. Business grade connections should be better.

That one is performing fine, no complaints.

It's variable and somewhat erratic, but clearly slower than I had before. It's not unusable like somebody stuck in the middle of nowhere with 1mbps over ADSL, but at present falls well short of what I expected.

Edited by charlestown (Mon 20-Mar-17 13:19:54)

In reply to a post by cloudscape:

Here is a test from another cct on the same interconnect in SYNC at 33M / 7.5M
http://www.thinkbroadband.com/speedtest/button/14900...

Hmmm.

Hi RobertoS,

He is not on a Business/Priority traffic product. He signed up for and has always been on a standard home product with a standard traffic/QOS profile.

So the single thread performance i.e. single download activity, reflecting performance things in the business world like a simple file transfer is what users of your service can expect?

If that is the case, then my recommendation is that the customer move to another provider who considered single thread performance important.

As a point of interest when was the last time your re-sync'ed, and have you ever left it off for 10-30 minutes? Though, that might not be necessary. I think on the ASUS you can close the PPP session without disconnecting the DSL? Try that if it can be done.

What I'm after at the moment is ending the PPP session for a few minutes, and possibly picking up different routing when you re-establish one. That could very well be what happened when they sorted out the initial problem, that you got onto a different entry point into the gateway(s).

We may be chasing a red herring with the wifi aspect. Though if your good experiences were on 2.4GHz, go back to it on the ASUS. Don't make both changes simultaneously, as if things do improve we won't know which cracked it.

Sorry, that last test from the other (33/7.5) circuit was done from a rather poor PC.

Repeated with a better performing PC:
http://www.thinkbroadband.com/speedtest/button/14900...

In reply to a post by charlestown:

I am actually on an unlimited business package that is more expensive than the home packages and supposedly free of congestion.

In reply to a post by cloudscape:

Hi RobertoS,

He is not on a Business/Priority traffic product. He signed up for and has always been on a standard home product with a standard traffic/QOS profile.

Interesting.

I've never left it off for that length of time. Last week they asked me not to touch any settings in the router until tomorrow, so I'm leaving it as requested.

At BT I had the HomeHub 5 and there was very little difference in speed between 5GHz and 2.4GHz. At the moment I can see 5GHz is faster anywhere in the house. Certain devices like my phone only recognise 2.4GHz.

In reply to a post by cloudscape:

Hi RobertoS,

He is not on a Business/Priority traffic product. He signed up for and has always been on a standard home product with a standard traffic/QOS profile.

Then why do I have a fixed IP as shown for your business packages? It's also clear from my early correspondence that I was signing up for the business package.

Hi,

We provide a fixed IP for all our circuits.

The difference with a business product is:

1/ Enhance Care Level
2/ Priority Traffic
3/ Guaranteed Fixed IP*

* Home users get a fixed IP, but this can be changed with prior notice if necessary.

On the 17th of February I requested a business package by email. I actually made a point of this because I wanted reliability. Evidently I have not received this.

I have checked our notes and I can see where the confusion arose. The customer queried our FUP asking if it was OK to run a business on a Home FTTC service.

He was told that that is fine and his account would be handled in line with our Business FUP, we did not insist that he purchased a business service.

Seeing 2017 in the date threw me. That email was actually on the 8th of February and not the 17th.

I never suggested I was being forced onto any package, but requested the business package with unlimited traffic after receiving an email from CloudScape outlining prices on the previous day.

Later in the same email I asked about the FUP, since it seemed to preclude anybody working from home who might need the connection for both domestic and business use, irrespective of which package was chosen. I was assured it wouldn't be a problem. From what I was told earlier today, it seems there would be no performance benefit from having the business package.

We can go back and forth like this all day long, but frankly I would rather be concentrating on my work, than losing time on this. For the first two days it was fine. Then my login was changed following 10 minutes of downtime and I've been troubleshooting ever since for over 2 weeks.

OK, just my 5 pence worth.........

Firstly my reasoning .... you say all was well and as expected until their login details were remotely changed by the CP since then it's all gone Pete Tong.

Suggestion, especially since a rep for the ISP is reading and posting in this thread ...... Ask them to completely delete your account from their remote access servers, strip it right out, every last bit of detail then have them rebuild, preferably with new log in details, a whole new account on their systems.

Doing this will hopefully resolve the issue, and if it doesn't you can be 100% sure it's not their end of things,

The change of username is a red herring.

When the customer went live our new interconnect was not yet active, so the new username just moved him onto the new pipe.

The problem is with WiFi.

I think enough has been said unless anyone can engage with him on how to improve his WiFi performance.

In reply to a post by cloudscape:

The change of username is a red herring.

When the customer went live our new interconnect was not yet active, so the new username just moved him onto the new pipe.

The problem is with WiFi.

I think enough has been said unless anyone can engage with him on how to improve his WiFi performance.

And yet there was no wifi issue a minute before that red herring was introduced.

Sorry, but how are you discounting my suggestion, especially when your reply states ....

When the customer went live our new interconnect was not yet active, so the new username just moved him onto the new pipe.

That would imply to me that the poster has only had issues since his circuit was shifted to your new 'interconnect' service.

Perhaps the ISP should try putting the customer onto a login that is on the old interconnect?

The speed tests run over Ethernet shown a distinct issue issues, and things like slow single thread issues feel worse once you add Wi-Fi to the mix.

What date was the new interconnect live? If might be interesting to look at all the speed tests we can see before that date and after that date.

Perhaps the ISP should try putting the customer onto a login that is on the old interconnect?

An even better suggestion than mine, since it moves the connection back 'as was'.

It is all too much of a coincidence to me that all was well until the ISP put the OP onto the 'new pipe'.

Edited by Zarjaz (Mon 20-Mar-17 19:31:16)

In reply to a post by cloudscape:

That one is performing fine, no complaints.

There is clear congestion throughout these speed tests.

The op has slow speeds over wired as well as wireless.

They may be blaming the wireless when they contact support, but all of us here can see the underperformance even using wired. With the best wireless in the world this issue would not be resolved.

It would appear you have the op on a congested link.

The rep probably doesn't realise that even if the user is "on a new under-contended interconnect on our network", that doesn't mean there cannot be congestion somewhere between the Openreach handover at the fibre headend exchange and where it reaches that interconnect.

For instance, if that interconnect were a shared BT Wholesale MSIL at a WBMC node, with Cloudscape buying plenty of capacity (a pipe) for their own customers, it could still be congested. Cloudscape could simply fail to get the throughput they expect. Even if their "interconnect" is one such.

That turned out to be a major problem for Plusnet a couple of years or so ago. Causing them to have to move to dedicated MSILs. Most of this Kitz section is relevant.

AIUI the same sort of troubles can arise on non-BT Wholesale backhaul networks.

Hi ukhardy07 and RobertoS

I am aware of both potential issues. We have already raised this with the carrier as we know there is no issue in our network or in the cross-connect between the carrier network and our BGP devices. We have had Openreach reset his profile and we have been running comparative tests on other circuits on the same interconnect, see this one which is in Sync at 33M / 7.5M http://www.thinkbroadband.com/speedtest/button/14900... and looks fine.

If there is an issue or congestion it is local to the OP or between him and the carrier backhaul. (FYI, in the 18 days he has been with us he has uploaded 272G and downloaded 175G)

I love the way people just assume we are ignoring the OP's concerns, this not the case as we have built a good reputation and intend to keep it.

A data dump

http://tbb.st/1490054073540724455
http://tbb.st/1490039554863030155
http://tbb.st/1490038452676435455
http://tbb.st/1490036271185606155
http://tbb.st/1490035746375830455
http://tbb.st/1490031034217715855
http://tbb.st/1490027428448145355
http://tbb.st/149001863926000733115
http://tbb.st/1490016909693260955
http://tbb.st/1490016492680889355
http://tbb.st/1490016398361202455
http://tbb.st/1490016275291207255
http://tbb.st/1490009722432291355
http://tbb.st/148999962522139848057

Various users, your test and the users ones are in there too

Hi Andrew,

Very useful. Is there anyway you can PM me the CLI's so we can see which, if any, of these has moved onto the new backhaul yet?

Regards

For some strange reason my cloud backup decided to start from scratch after moving to the new pipe. I have no idea why that happened, since everything had already been backed up previously, but that would have been 250+ gigs. That download figure seems quite a bit higher than I would expect, though I imagine a large part of that might be Amazon Prime. I did request unlimited bandwidth.

Regarding the point of congestion, it's clear that my line becomes really slow at peak times and faster late at night, whether on Ethernet or wifi. At 1.00am I might be getting 25mbps on wireless and at 8.00am I'm lucky to get 7mbps.

Caller Line ID? You have run the test are you not aware that we don't ask people for that.

User privacy is important, so if there is a way to ID if people are on the new backhaul via IP address we can do that filter for you, but do not normally share IP address level information.

Hi Andrew,

IP's would be great.

Regards

PM contains two links to speed test versions where we will be able to share a little more information and help get to the bottom of whether there is any difference between how the old and new are behaving

Will leave sharing the links to the ISP, since the idea is one group use one version, and the others the other to make identifying the differences easier

Was that backup running whilst you were doing speed tests?

No. I made sure nothing in the house was running, including other devices. The backups happen once an hour, but they can be manually paused.

thanks Andrew, Much appreciated

If it had been running as its a backup would have expected the upload results to be impacted too

This is just a quick update to say that Cloudscape have been in touch and the problem now appears to have been resolved. I'll leave it for them to provide the technical explanation, but it wasn't anything at my end.

Yesterday we ran a test here via Ethernet after rebooting the router that showed speeds were significantly down on expected results and at one point wireless slumped to just 3mbps at around 6.00pm. It suddenly improved quite dramatically at around 9.00pm last night and by midnight it was nudging 60mbps.

Since rebooting the router yesterday the sync rate has dropped to 66mbps, however it's still in the DLM period so that can be looked at over the next few days. The actual download speed with wireless was tested a few minutes ago and I was seeing 62mbps. http://www.speedtest.net/my-result/6152087498

A huge thank you to everybody here who helped to resolve this.

There isn't a DLM period. DLM is always running - there is no 10 day training on FTTC. DLM will adjust the line as needed for the entire time you have it. If it is reducing then it is probably because of noise (possibly from cross talk). That could change over time and depending on line conditions things like G.INP may kick in.

I've always heard contradictory information on that point regarding a DLM period for fibre, so never known for sure what is right. It should therefore speed up itself over the next few days, since I'll leave it switched on and not touch anything.

ISPs still state it but it isn't true. There is nothing that happens in the first 10 days that doesn't happen during the whole time the line exists. Generally the line will connect as fast as possible when it is first plugged in - and then for the most part it will, if anything, slow down over time. This is DLM adjusting to the conditions it finds during normal running.

However, G.INP if triggered could result in an increase in speed. Other technologies BT are doing could also help (like reducing the noise margin). But, if there is no outside changes then DLM is generally a downward curve.

It used to sync at around 78mbps at this address. We'll see over the next few days how it goes and whether it adjusts itself upward.

The various fiddling may have not helped, but the DLM system generally takes weeks to relent rather than days, so just get on with life rather than watching it expecting it to change suddenly.

Not looked at the test data but will do at some point.

Sure I'm not worried at all while it's like this. Any delays or slowness are more likely owed to badly designed websites or overcrowding on cheap shared web hosts.

It was a real headache when the line kept dropping and speeds were sometimes under 6mbps, but any improvement above the current speed is fairly academic in real terms.

Hi All,

Thanks for the input from everyone particularly Andrew for help with the speed tests and RobertoS for his useful comments.

With the help of our back-haul provider and our cross-link provider in Telehouse we found a fibre link that was incorrectly rate capped, this was then uncapped last night. As this was not in our internal network it did not affect every one, but knowing who was affected helped with the resolution.

So not 'his wifi' then ?

Thank you for coming back and providing clear information on what the actual problem was.

Hi Zarjaz

The initial symptoms where slow WiFi and when we connected to his PC with RDC on a wired connection we could not spot any issue. It was only when his wired connectivity also degraded that we traced the underlying issue.

At least you did an investigation and got it sorted!
Better than most providers

We could all see the underperformance over here. I can see how given the user discussing the wifi it would easily be misunderstood.