Kitz Forum

Broadband Related => FTTC and FTTP Issues => Topic started by: spudgun on June 28, 2019, 03:20:06 PM

Title: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: spudgun on June 28, 2019, 03:20:06 PM
Good afternoon all,

This isn't a problem as such, but i've raised this so that I can add to my own knowledge and understand my line and G.Fast technology a little better.

I've had a visit from a fantastic and helpful Openreach engineer this week in relation to my G.Fast circuit.

According to his measurement equipment (see pic below) that was plugged in to my line, it is capable of 253mb or thereabouts.

(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fi68.tinypic.com%2F11izshu.jpg&hash=4d02922462a84437ce89a1c7c54e3d8d)

This is broadly similar to the maximum observed rate that is noted for my line via the Openreach database (see image below) - please note that the engineer reading and max observed rate are both on the same day.

(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fi67.tinypic.com%2F2gv10ec.jpg&hash=db9b5abed257f6db5667ebbf85125745)

There is a difference between the sync speed that is reported on my ISP's portal, however, (see image below)  The measured throughput on my line (setup is a MT992 + AC68-U running latest Merlin) is consistent with the sync speeds that my ISP is reporting when overheads are taken into account.

(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fi66.tinypic.com%2F30i9ibo.jpg&hash=d3d7d3ff1310236ffd0b23f6ff372589)

Does anyone know why there might be such a big difference between Openreach's information and that from my ISP?

I look forward to seeing what the experts on here make of this :)
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: burakkucat on June 28, 2019, 03:45:51 PM
I will offer the opinion that any measurements made by Openreach will show what the circuit is capable of achieving without considering any overheads; any service provider's profile modifying/adjusting/tweaking/banding the result.

The Openreach G.Fast Closeout Test reports: 253 Mbps DS / 30 Mbps US
The BT Wholesale checker reports:           247 Mbps DS / 30 Mbps US
Your ISP (is that Zen?) reports:            229 Mbps DS / 30 Mbps US

(All rounded to the nearest whole number Mbps.)

To me, that table of results looks reasonable. But I may have missed something significant. So I shall watch for comments from other members.  :)

[Edited to correct the error.]
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: spudgun on June 28, 2019, 03:49:47 PM
Thank you for your input, that was my first thought too - but probably not  wise by openreach to have that on their website as some people might expect to get the 'no overheads' speed and raise issues.

I can confirm that Zen is the ISP and your summary is accurate (apart from the Zen upstream that is 30mb in line with the other two).
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: burakkucat on June 28, 2019, 03:55:33 PM
. . . probably not  wise by openreach to have that on their website . . .

To me, that looks like the output from the BT Wholesale Broadband checker.

Quote
. . . (apart from the Zen upstream that is 30mb in line with the other two).

Post edited to correct my error.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: ejs on June 28, 2019, 03:58:31 PM
Perhaps it's simply that the Openreach tester is the better modem.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: kitz on June 29, 2019, 02:16:15 AM
^ Possible!
Without any stats from the MT992 we are working blind.   :(

Also there is the possibility that DLM could have changed something which affects the sync speed.  I'm not sure if recent FTTC DLM amendments also follow over to g.fast or not, as g.fast wasn't mentioned.   
So I suppose there is a chance that the line was on full open profile when the engineer performed his tests but after a day the default DLM profile has kicked in - whatever that is for g.fast?     It was last year when the engineer ran tests on my daughters line but I know he reset DLM to Open yet a few days later they had interleaving applied.   

You may be able to spot interleaving based on latency - however that could depend where you live and which Zen gateway you are going through.   
If you are on their Manchester PoP and live down South it could be harder to tell.   With my daughter living up north, being on their Manchester PoP makes no difference to their overall latency.

----

PS - For clarification:

Zen are seeing you sync at 236737.
229151 is your IPprofile
TR101 basically just means Ethernet backhaul (21CN WBC/WMBC).
Interleave Auto - means can be on or off as dictated by DLM.  This is historic from the days of ADSL as it can't be set as off when on VDSL

You many be able to get Zen to give you your DLM profile if they run a GEA test - although it does sometimes lag behind by 2 weeks.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: adslmax on June 29, 2019, 08:47:23 AM
There seem to be Pro's & Con's on gfast there. Don't understand why BT Openreach tester get a better result than BTw estimated and Zen profile on it.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: kitz on June 29, 2019, 09:27:41 AM
Quote
Don't understand why BT Openreach tester get a better result than BTw estimated and Zen profile on it.

If you read the rest of the thread, there's several reasons given as to why this could be :/
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: waltergmw on June 30, 2019, 10:47:02 AM
For those who haven't looked recently the BT Openreach data (which often used the BT W "Impacted" figures) has now been withdrawn so Openreach now only provide some cabinet location details.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: Chrysalis on June 30, 2019, 04:16:02 PM
The BTw estimates are most definitely sync speed (before overheads), not throughput.

Openreach testing kit the same.

The reason been both entities are not responsible for throughput.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: spudgun on June 30, 2019, 09:14:29 PM
Apologies for the delay in responding to all of the kind and helpful points made above, but I've been away for the weekend.

In response to the points raised, I am in Gloucestershire, and have previously raised a ticket with Zen to ask to have my routing not go via Manchester to London as that makes no geographic sense and they have taken care of that. Whilst I was going via Manchester pings were about 16ms to bbc.co.uk, but with London routing this is now either 9 or 10ms. My estimates of distance travelled against speed make me believe that this connection isn't interleaved, but I will stand corrected if the experts here can find fault in my logic.

I agree that the lack of stats from the MT992 make understanding G.Fast difficult and that is one of the reasons that I raised this post as sharing knowledge/experience is perhaps the best way that we can build up a limited understanding.

I did get a look at a separate screen on the device that the engineer was using that gave the snr info for my sync (sadly no photo of this) and that was 3db up and down - which i understand to be the default for g.fast - and the engineer believed that to be the case too.

The difference between the testing device and BT wholesale information is made a little bit more strange as I spoke to Zen on Friday night before I left and they said my line was synced at 247mb (which matches the BT wholesale observation) - but the line data that is available on their own pages contradicts this (posted in the OP) and the max throughput that I am seeing matches exactly the downstream of 229 (sync of 236).

Don't get me wrong, I'm not complaining or anything here for a line length of 310m (again measured by the engineer's kit) I think i've done really well and the line was really stable until all of the underground chambers flooded about 2 weeks ago and knocked everything out (including making voice calls very difficult - and this is why the engineer visited!), but I am actively interested in technology (as many here probably are), so thought this might be worthy of discussion.

Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: re0 on June 30, 2019, 09:50:12 PM
I had to do the same on Zen in regards to routing. Initially, they were reluctant and refusing to change my routing via email (and maybe telephone). However, I had raised a fault for low single-thread speeds and as soon as I mentioned to them about the routing, they changed it to London from Manchester since apparently it was a mistake. Not only did this reduce my latency to as low as 9-10ms, but also fixed the single-thread speed issues.

There aren't that many places in Gloucestershire that have G.fast. Cheltenham has had it for a while, and the pods have been around there for some time. I haven't noticed any further popping-ups of pods in other cities and towns personally.

If you're interested in stats, I would recommend the XMG3927-B50A. You can find the stats in the UI and also use SSH/telnet, and it is possible to use DSL stats with (as I do, with the Raspberry Pi).
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: j0hn on June 30, 2019, 09:52:45 PM
The default target SNR is 3dB but the DLM is considerably more flexible on G.Fast and can go up/down in 1dB increments.
I've seen it go as high as 15dB.

I'd be surprised if the line was as long as 310m.

What was the attenuation on VDSL2?
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: spudgun on June 30, 2019, 10:10:27 PM
I'm really sorry, but I can't remember what the attenuation was on VDSL and I don't have any stats logs saved. I was on the full 80/20 which dipped to 76/20 about a month before G.Inp was launched, after which the full 80/20 was back.

The engineer's kit definitely said 310m as I remember that most clearly from when he was here and my best guess is that is about 50m more than a direct line route (based on nothing more than me pacing it out) so I have no reason to doubt its accuracy.

I think I've read on here that A+A will sell you one of those Zyxel boxes - any idea how much we are talking about as I might think about it next pay day?
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: burakkucat on June 30, 2019, 10:44:24 PM
I think I've read on here that A+A will sell you one of those Zyxel boxes - any idea how much we are talking about as I might think about it next pay day?

Rumoured to be in the  order of £100 - £150 (ex.VAT), £120 - £180 (inc. VAT) :-X

[Edited to account for VAT.]
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: kitz on June 30, 2019, 11:44:26 PM
If so, seems fair enough.   

I paid £120 for my VMG8324 because it was the first decent VDSL2 modem/router that was under £200.  I took a risk because it was brand new to the UK and I've not regretted a single penny of it, despite they are now much cheaper.

Bearing in mind choice is very limited for g.fast modems and that the ASUS DSL-AC88U is £200+  then even if the Zyxel is £150 it still seems a decent option.  When the market has little competition then prices do tend to be steep - look at the stupid prices the Huawei MT992 can command on ebay. :(
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: re0 on July 01, 2019, 12:06:39 AM
The XMG3927-B50A cost me a little under £150 from A&A, but would be unable to say what the price would be now - may be worth enquiring.

The DSL-AC88U has a slightly higher WiFi spec at AC3100, versus AC2400 on the XMG3927-B50A - although given the price and functionality required, it will be down to your judgement which is likely better. I haven't tried the former, only the latter. Can't say I have had any significant issues with the device as a router, and the modem side of things has been really solid.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: spudgun on July 01, 2019, 08:36:16 AM
Thank you both for that, I will consider it at the end of July when I get paid. I have experience with VMG8324 and VMG8924 on my line when it had FTTC - but neither performed as well as a HG612 in terms of connection speed and reliability and I prefered the routing options that were available on my trusty AC68U - although it might be tempting to try a single box on G.fast to get access to stats.

Out of interest, for those who aren't on max sync on G.fast - is there a noticable difference between the max observed downstream on the wholesale checker http://www.dslchecker.bt.com/and what you can see either via your ISP or modem, or even what your best guess is from speed testing your line?

Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: adslmax on July 17, 2019, 07:20:52 PM
If you're interested in stats, I would recommend the XMG3927-B50A. You can find the stats in the UI and also use SSH/telnet, and it is possible to use DSL stats with (as I do, with the Raspberry Pi).

I like to know this too. Any instruction how to use my XMG3927-B50A work with dslstats (for FTTC connected)
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: re0 on July 17, 2019, 09:36:13 PM
DSLStats just needs to be configured to use the xdslctl command, and it will be fine.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: fat jez on July 29, 2019, 11:07:30 PM
I was going to start my own thread about my G.fast experience so far, but Spudgunís experiences seem similar to mine.

I had G.fast enabled last Thursday, the 25th July.  The OR engineer replaced my socket and left me with a Huawei MT992 modem. He was a good guy and we had a good chat (Iím an ex mobile telecoms engineer).  One of the things he showed me was his test equipment plugged into my socket, showing a download sync of just over 350 and an upload in the mid to high 50s. I didnít get the SNR. My line length is just under 180m (that came from him as well).

This ties in with the ADSL Wholesale page. (See first attachment), which shows 330 down and 50 up.

However, my line speed is just over 300Mbps (see second attachment), which makes me wonder if I have a stuck IP profile? Iíd have expected to get about an extra 16Mbps on my line. Am I just being impatient and need to give things time to stabilise?

Like Spudgun, I find the lack of line stats to be frustrating!
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: re0 on July 29, 2019, 11:51:11 PM
One of the things he showed me was his test equipment plugged into my socket, showing a download sync of just over 350 and an upload in the mid to high 50s.
Those are attainable rates. Line will sync in line with the profile - 330/50 Mbps max.

However, my line speed is just over 300Mbps (see second attachment), which makes me wonder if I have a stuck IP profile? Iíd have expected to get about an extra 16Mbps on my line. Am I just being impatient and need to give things time to stabilise?
I'm with Zen, also sync'd at the full 330/50. I get approx. 293 Mbps downstream in speedtests. Line data shows sync rate at approx 313 Mbps. Line rate data hasn't updated sinc earlier this month, though I don't think either of us have a stuck profile; I don't think things will change for you.

Like Spudgun, I find the lack of line stats to be frustrating!
You do have the option of buying your own hardware! :) Zyxel XMG3927-B50A certainly recommended.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: adslmax on July 30, 2019, 05:53:06 AM
Like Spudgun, I find the lack of line stats to be frustrating!

First of all, congrats to your G.fast 330/50 as for Zyxel XMG3927-B50A are certainly recommended and ask AAISP to send you one but you have to pay £153.98 inc P&P with Royal Mail Tracked 24 and this will work with DSLstats with xdslctl command config.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: fat jez on July 30, 2019, 08:42:07 AM
Those are attainable rates. Line will sync in line with the profile - 330/50 Mbps max.

Yes, I know. :) I was trying to indicate they my line sync should be at the maximum 330/50 because there's some headroom. I'm the only G.fast user on my cabinet, or perhaps more accurately, I was the first G.fast user on my cabinet, so there shouldn't be any issues with interference (although it should have vectoring to take care of this anyway).

I'm with Zen, also sync'd at the full 330/50. I get approx. 293 Mbps downstream in speedtests. Line data shows sync rate at approx 313 Mbps. Line rate data hasn't updated sinc earlier this month, though I don't think either of us have a stuck profile; I don't think things will change for you.

Your speedtest is in line with mine. I was expecting my IP Profile would be around 319.44, (96.4% of 330), rather than the 303077 Kbps that Zen shows. They also indicate my sync rate is 313110, rather than 330000, which is what prompted my original question.



You do have the option of buying your own hardware! :) Zyxel XMG3927-B50A certainly recommended.

I know. But G.fast modems (I don't need a router, I have a Netgear Orbi wireless setup) are as rare as hen's teeth. Ones that do stats are even rarer! I'm holding out hope for the Draytek V166.

Thanks for your responses, it does seem as though my line is in the right ball park area, performance wise.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: fat jez on July 30, 2019, 08:45:17 AM
First of all, congrats to your G.fast 330/50 as for Zyxel XMG3927-B50A are certainly recommended and ask AAISP to send you one but you have to pay £153.98 inc P&P with Royal Mail Tracked 24 and this will work with DSLstats with xdslctl command config.

Thanks Max :) £153.98 is a bit more than I'd want to pay for a modem at the moment, especially one that doesn't yet have a proven track record (i.e. I can't find much about it). I'm hoping the Draytek V166 will be more widely available and also support line stats.

*edit* I see why the Zyxel is so pricey, itís a router, not a modem.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: j0hn on July 30, 2019, 11:52:32 AM
Perhaps G.Fast defaults to Retransmission High, which would give an IP profile nearer 91%.
That would fit more with both your speed tests.

No idea if the stats on the Zyxel show bearer 0 info with G.Fast.

Also need to take in to account that the engineers JDSU syncs higher than the Huawei modem.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: re0 on July 30, 2019, 12:36:16 PM
Bearer 0 from the Zyxel:
Code: [Select]
                        Bearer 0
INP:            550.00          535.00
INPRein:        0.00            0.00
delay:          0               0
PER:            0.00            467.44
OR:             755.48          297.33
AgR:            34526.15        26598.55

I would have thought that the higher INP level necessary for G.fast would have been at least part of the reason. No INPRein here though.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: fat jez on July 30, 2019, 12:43:11 PM
Also need to take in to account that the engineers JDSU syncs higher than the Huawei modem.

I didn't know that! :) He was syncing at ~350Mbps and the ADSL Wholesale checker shows me at 330/50, can these figures be trusted? i.e. am I actually syncing at 330/50? My upload speed tests at around 48Mbps suggest that the upload figure at least is accurate.

Without line stats I'm blind and I don't want to waste money buying a router when I only need a modem.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: re0 on July 30, 2019, 02:09:19 PM
He was syncing at ~350Mbps and the ADSL Wholesale checker shows me at 330/50, can these figures be trusted? i.e. am I actually syncing at 330/50? My upload speed tests at around 48Mbps suggest that the upload figure at least is accurate.
If his JDSU was sync'd at 330 with a nice bit of overhead with max rate of ~350 Mbps, your modem should not have much trouble maintaining 330 if it's a clean line with not much interference. I have found the DSL checker to be reliable in the sync rate results it returns, though the downside is that the data can be a few days old. 48 Mbps upstream is about what I get, too.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: fat jez on July 30, 2019, 10:09:05 PM
re0, you appear to have line stats. can you see your sync speed? The reason I ask, is the line data on Zen's customer portal says that the BRAS profiles is listed in the Downstream column and is 88% of the current sync rate.  That would put my sync rate at 354594 (figure from download column divided by 0.88).

How does your download figure compare to 88% of your sync speed?
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: re0 on July 30, 2019, 10:49:44 PM
Code: [Select]
Max:    Upstream rate = 66064 Kbps, Downstream rate = 345788 Kbps
Bearer: 0, Upstream rate = 49871 Kbps, Downstream rate = 330011 Kbps

303139 (BRAS) / 0.88 = 344,476 (rounded)
330011 (sync (modem)) * 0.88 = 290,410 (rounded)
313174 (sync (user portal)) * 0.88 = 275,593 (rounded)
313174 (sync (user portal)) / 0.88 = 355,880 (rounded)

I have a feeling it is closer to 91-92% as opposed to 88%, since the first two calculations seem to make sense using that. I actually don't know the answer, but I would not be too surprised if G.fast needed a larger overhead for high INP.
Title: Re: G.Fast - Understanding a difference of information between ISP and Openreach
Post by: fat jez on July 30, 2019, 11:13:39 PM
Thanks for this. Makes you wonder if 330Mbps was chosen as the sync rate because it gives around 300Mbps after potential overheads?