80/20 FTTC Trial Open

[#1]

@mikeyriot It only worked once, subsequent attempts have returned:
Looks like BT refused to test, possibly because a speed test or BRAS check has been run recently which prevents this tool from working. Try again in a few hours.
Debug: BT ISE
More than an hour apart, so it's not the one hour restriction. I'll try again tomorrow, when I will probably be back with more about my speed issues. OK, it's still fast enough for all practical purposes and faster than I was getting before (much faster for the uploads) but it's frustrating to get such good performance for a short while and then for it to take a dive for no apparent reason.

[#2]

My connection has slowed down this evening, to be honest I kind of expected it due to being a Friday night and a busy time for people using the internet.
Its at between 50 and 55meg at the moment, it started slowing down around 7pm, I fully expect it to be back to normal by the late evening.  And if thats the slowest it's gonna get then I am very happy.

[#3]

To sum up the evening, the speeds dropped from 75meg down to 50meg between 7 and 9pm, they then dropped a bit more between 9pm and 11pm to around 45meg and by midnight I was back up to 75meg.

I am not sure if I really need to post this but as I am part of the trial I believe it to be useful information. Obviously peak times on a Friday night play a big part in speed as I lost around 30meg. I don't mind this as its still faster than the original 40meg service at the worst of times.

[#4]

Did you notice any slow-down of the original 40meg service at similar times?

[#5]

Quote from: paulpicks21

To sum up the evening, the speeds dropped from 75meg down to 50meg between 7 and 9pm, they then dropped a bit more between 9pm and 11pm to around 45meg and by midnight I was back up to 75meg.

My problem is a little different, although my IP profile has dropped by about 7Mb, which is annoying, but not the end of the world, since yesterday I am being "capped" to at least 20Mbps below this at all times, not just peak times. I am still hopeful that all of this will magically sort itself out, and I know it's only a trial that is currently not costing me anything but ...

[#6]

Quote from: walker23

Did you notice any slow-down of the original 40meg service at similar times?

I have had the Pro add on since I first got fttc around a month ago, the speed was constant at 37.5 meg no matter what time of day or night.

[#7]

Looks like my 80/20 connection has been stable for the last couple of days. BT IP profile has remained at 67.53/20, and speedtest.net throughput is 61-65 / 14.5-15.5.
This is with the Netgear WNR1000v3 router, using a wired connection to a newish PC. I'm going to pick up the new one from the P.O. today. With the IP profile I have, I doubt whether the gigabit link from the ZyXEL will make any difference, but we'll see.
Graham
(paulpicks21 - some of your increased usage might be due to speed tests. My usage certainly went up a lot on the first day, and the only thing I was doing much more was running speed tests.)

[#8]

Quote from: KevinG

Oh dear. I just unlocked my modem (got a replacement last week so hadn't done it yet), which obviously meant powering it off for a short while, and my line rate is showing as 67737, which must be lower than it was as I had a profile of 72.89M and was getting speed test results of around 70M.

You'd have been just over a couple of days into the trial then, I think, which would be about the time that DLM would have made its mind up. I suspect it has turned some interleaving on, and the extra parity information being carried for FEC has reduced your line speed a little.
When you do a "--show", it tells you a few parameters that (I think) describe the interleaving configuration:
- INP tells you how much "Impulse Noise Protection" DLM has decided it needs
- Delay seems to show how much additional delay is to be added by the interleaving scheme that is chosen
- R shows how many parity bytes get added to each RS block. If zero, then interleaving is off
- N shows how big each RS block is.
I currently have these values, with no interleaving:
- INP = 0.00
- Delay = 0.00
- R = 0
- N = 240
The line used to not need any interleaving, but had some anyway, at a light level:
- INP = 0.00
- Delay = 1.00
- R = 14
- N = 255
So around 5% of the line capacity was taken up for the parity bytes for FEC. If I remember correctly, it had this from day 1 (ie before DLM made any changes).

[#9]

Quote from: tomm

Quote from: WWWombat What I did notice is that speeds can be variable, and are now dependent on the machine running the test. For example: my old laptop (2006 vintage) can only sustain 65Mbps on speedtest.net As the speeds get faster dunno what's on that laptop but WinXP in particular can benefit from the tuning of the TCP receive window. You can tune stuff using the tweaktest and drtcp tools from here. http://www.dslreports.com/tools Tom.

Thanks Tom - it was XP, and was indeed related to the receive window. I was already trying out some experiments based on a post on the TBB "fibre" forum.
I decided it was worth posting in a thread of its own, so it didn't get lost in the mass of posts in here.

[#10]

Quote from: WWWombat

You'd have been just over a couple of days into the trial then, I think, which would be about the time that DLM would have made its mind up. I suspect it has turned some interleaving on, and the extra parity information being carried for FEC has reduced your line speed a little.

Annoying, but liveable with if I weren't having another 20Mbps docked off for no apparent reason!

Quote from: WWWombat

When you do a "--show", it tells you a few parameters that (I think) describe the interleaving configuration: - INP tells you how much "Impulse Noise Protection" DLM has decided it needs - Delay seems to show how much additional delay is to be added by the interleaving scheme that is chosen - R shows how many parity bytes get added to each RS block. If zero, then interleaving is off - N shows how big each RS block is. I currently have these values, with no interleaving: - INP = 0.00 - Delay = 0.00 - R = 0 - N = 240 The line used to not need any interleaving, but had some anyway, at a light level: - INP = 0.00 - Delay = 1.00 - R = 14 - N = 255 So around 5% of the line capacity was taken up for the parity bytes for FEC. If I remember correctly, it had this from day 1 (ie before DLM made any changes).

Is this the relevant bit? Couldn't see how to get exactly the parameters you quoted.

INP:            3.00            0.00
PER:            1.45            4.27
delay:          8.00            0.00
OR:             132.35          59.93

[#11]

Hi all,
My 80/20 trial burst into life and I'm a very happy customer. Incidently I'm 600 metres away from the FTTC cabinet.
It's hard to believe that just two year ago I moved my adsl router into the garage adjacent to the NTE5 to try and get 2Mbps...

Before the 80/20 upgrade my speed was about 34Mbps
I've also noticed that my old faithful XP machine is no longer up to the job and is too slow to report speedtest results. The results above were from a newer Vista PC.
Jon

[#12]

Quote from: Wicken

I've also noticed that my old faithful XP machine is no longer up to the job and is too slow to report speedtest results. The results above were from a newer Vista PC.

Have you increased RWIN on your XP computer as described in the 80/20 Speeds - Windows XP limitations topic?
The isn't necessary on Vista (or W7) since it adjusts dynamically to suit the traffic.

[#13]

Quote from: KevinG

Annoying, but liveable with if I weren't having another 20Mbps docked off for no apparent reason!

I agree with that - but I do wonder where that "I can live with that" threshold is, where people decide it is worth paying extra over the 40/10 service.
Hopefully Dave or Adam is on the case of the 20Mbps - that seems a legitimate issue for BT to sort out.

Quote

Is this the relevant bit? Couldn't see how to get exactly the parameters you quoted. INP:            3.00            0.00 PER:            1.45            4.27 delay:          8.00            0.00 OR:             132.35          59.93

That gives the first two bits, which are the "input" into interleaving. The INP value tells the DSLAM that it needs to protect against 3ms noise bursts, while the "delay" value (I think) says that interleaving can add 8ms of delay onto the latency.
The other two parameters "R" and "N" come from a block earlier in the output, labelled "VDSL2 framing". They'll give us an idea of just how much overhead is being stolen for FEC - but I'd expect it is a fair amount.

[#14]

Quote from: WWWombat

I agree with that - but I do wonder where that "I can live with that" threshold is, where people decide it is worth paying extra over the 40/10 service.

How about a 40/20 service priced mid way between? I'd find that far more useful (if fibre was available in the first place).

[#15]

Quote from: WWWombat

where people decide it is worth paying extra over the 40/10 service.

Or, strictly speaking, over the 40/2 service (at present).