Kitz Forum
Announcements => News Articles => Topic started by: Bowdon on April 16, 2018, 06:15:03 PM
-
https://www.ispreview.co.uk/index.php/2018/04/openreach-to-test-sra-on-uk-fttc-superfast-broadband-lines-again.html (https://www.ispreview.co.uk/index.php/2018/04/openreach-to-test-sra-on-uk-fttc-superfast-broadband-lines-again.html)
Openreach (BT) has announced that they intend to conduct a new Proof of Concept (PoC) to test the impact of applying Seamless Rate Adaption (SRA) to their 40-80Mbps capable VDSL2 based Fibre-to-the-Cabinet (FTTC) broadband lines, which could help to keep such lines stable.
At this point anybody familiar with SRA will possibly raise a wry smile, after all it’s not the first time that Openreach have tried to adopt it and indeed it’s already been implemented on their new hybrid-fibre G.fast service. The operator also tried to adopt it as part of their Long Reach VDSL (LR-VDSL) trial but that has now been shelved.
Prior to that the last time that we saw SRA being attempted on Openreach’s hybrid fibre VDSL2 tech was as part of their Vectoring trials between 2014 and 2015 (vectoring is designed to remove crosstalk interference and thus improve performance), which ran into a fair few development challenges and only ever resulted in a highly targeted deployment of Vectoring (the SRA component of that trial was not a big problem).
UPDATE 5:19pm
We understand from ISPs that the trial will take place on just 200 lines via Huawei cabinets, which will start around mid-May and run until the end of June 2018. Apparently the trial will mostly involve Openreach’s own employees and their Network Operations Centre will then apply SRA line by line (Dynamic Line Management (DLM) will also be disabled).
Assuming all goes well then the prospect exists for a larger trial of around 100,000 lines, which could begin around September and run until the end of this year.
I had to post this here as I couldn't find the original news story.
-
I notice it says Huawei cabinets, no surprise there. Can ECI actually support SRA?
-
If the ECI do not support SRA they need to rip them all out and put in Huawei.
What an absolute kick in the teeth that yet again the ECI cabs are ignored.
How about testing/trialing on ECI first? Especially as ECI can't get Ginp, vectoring or 3db or anything else?
They really are utter sh*t and should be returned to ECI for a full refund and appropriate compensation.
FFS instead of Gfast replace the ECI cabinets with Huawei with everything turned on. It will be cheaper, and give more benefit to a greater number of people.
Oh I forgot they can charge twice as much for Gfast and hopefully force users on ECI to upgrade as they as so cr*p compared to Huawei.
Rant over.
-
I guess we know where the 'continued effort on resolving the G.INP issues with ECI' is currently going, to this new trial for Huawei only.
-
I saw that this morning on the Openreach site. I didn't know it was just Huaweis though. :(
pffffft the divide just gets bigger. :'(
I don't see any way out of this for those on ECI cabs. G.fast wont do anything for me and any others who are too far from the cab to benefit.
-
They really are utter sh*t and should be returned to ECI for a full refund and appropriate compensation.
As an aside do people realise that ECI and the V41 goes down in history as the first ever system based vectoring DSLAM in the world -linky (https://books.google.co.uk/books?id=dnbPoJfkIVkC&pg=PA297&lpg=PA297&dq=system+based+vectoring+eci&source=bl&ots=xnajoPMPJA&sig=Un1V0q8uICXak9TngpMqe1U60ak&hl=en&sa=X&ved=0ahUKEwj-iamC-r_aAhWGW8AKHac5AV4Q6AEIXjAI#v=onepage&q=system%20based%20vectoring%20eci&f=false)
Although the theory of vectoring had been around for a while, it was all done on the linecard, which wasn't always popular because of the in-efficiency of wastage of ports.
From what I can gather the V41 didn't become commercially available until either very late 2011/early 2012.
So when BT/Openreach sat down in ~2010 and started designing their network there was no such thing as vectored DSLAMs. By the time the V41's came out then Openreach would have already had on order several hundred M41s.
Although I have mentioned this in the past about how close timescales were with the new V41's, I more recently forgot about problems of lead in time until reading an NBN article about broadband trends in Australia (https://broadbandtrends.wordpress.com/category/vdsl2-vectoring/) and the difficulties of planning a network during a time of emerging new technologies. NBN started planning in 2010 for initial rolloug of 2012-2014 (just slightly later than Openreach). Those difficulties are :-
1) Time to Market
2) Quote "at the time of the initial planning for NBN – technologies such as DOCSIS 3.1, VDSL2 Vectoring and G.fast did not exist"
So here we are in 2018 saying why didn't Openreach buy V-41's when placing their orders in 2011. Perhaps they may have done if there was such a thing as system level vectoring and V-41's actually available when they started designing their network. It explains why Deutsch Telekom and France also rolled out FTTC with M-41's. I guess you can't plan and order something which isn't yet available. :'(
-
It will be intersting to see the results of this trial, who will benefit from it most and whether it will actually make it out of the trial and into a nationwide rollout. I imagine it could benefit almost everyone, but especially those that see huge variations in background conditions. Though it is certainly a shame that there is not much hope for people on ECI kit.
SRA has been virtually absent in xDSL products rolled out by BT/Openreach (except for some trials which have existed in more recent years), though some LLU providers back on ADSL2+ did activate SRA on their services. If I recall correctly ADSL24 LLU had SRA, along with some pointers that would imply that O2/Be trialled the capabilities on their own LLU hardware. Back during the rollout of BT's 21CN WBC services to exchanges, it seems that BT chose not to implement SRA as they confirmed so during the later stage of the rollout in 2011 (https://www.ispreview.co.uk/story/2011/09/22/bt-not-implementing-adsl2-broadband-seamless-rate-adaptation-technology.html (https://www.ispreview.co.uk/story/2011/09/22/bt-not-implementing-adsl2-broadband-seamless-rate-adaptation-technology.html)). So this could be the firist emergence of SRA on Openreach's network with availability to all ISPs on the platform if the trial is successful.
-
Would SRA significantly affect the need for DLM and all of its complexities/foibles? Your line gets worse... SRA drops the speed. Your line gets better and SRA increases it again? No more capping/banding?
-
Would SRA significantly affect the need for DLM and all of its complexities/foibles? Your line gets worse... SRA drops the speed. Your line gets better and SRA increases it again? No more capping/banding?
According to the information in the main article linked in this topic, G.fast has SRA. However, according to https://www.btplc.com/SINet/sins/pdf/STIN520v2p0.pdf (https://www.btplc.com/SINet/sins/pdf/STIN520v2p0.pdf) it still also uses DLM. I cannot say this will be representive of whether it will impact the need for it in VDSL (FTTC) connections. But I would imagine that there will always be an underlying DLM that caps those extremely problematic lines, and paremeters such as the SNRM and retransmission will continue to be adjusted as per the line's performance since SRA can be a good combat against slowly changing line conditions (steady increase of errors, more intererence) but immediate changes may probably still cause drops.
-
I suspect that the reason DLM has been turned off for the lines on the trial (if I read it correctly) is because DLM might see fluctuating sync rates as a sign of instability and therefore band a line (like my first line is currently banded). It's possible that if SRA was rolled out that banding may no longer be used for such a situation, but remain for issues such as rapid re-syncs.
-
Hi
Here we go again, another trial of a technology that Openreach should have deployed and tested with at the very start, seamless rate adaption (SRA) raises it's head again this time on VDSL. :lol: A small trial involving staff to start with, news posted at ISPreview (http://"https://www.ispreview.co.uk/index.php/2018/04/openreach-to-test-sra-on-uk-fttc-superfast-broadband-lines-again.html").
No surprise this looks like a Huawei only trial, likely never to come to an ECI cabinet.
Regards
Phil
-
There's already a topic for it in "News Articles" at http://forum.kitz.co.uk/index.php/topic,21373.0.html (http://forum.kitz.co.uk/index.php/topic,21373.0.html).
Though I do truly feel sorry for those on ECI cabinets. :lol:
-
Hi
I had SRA for short time whilst on a LLU ADSL line.
I did work, with the sync speed fluctuating quite often without a full resync and the SNR staying pretty constant, however it didn't make the line any more stable overall or more more immune from resyncs. I guess the problem with some lines is there are times where there is more noise than a 6db margin can cope with, so it makes no difference if SRA is keeping that margin constant or not, it's going to get knocked out regardless. It was more annoying really as sync speed would drop at night, so it was trading a fluctuating SNR margin for a fluctuating sync speed, but providing no more immunity to resyncs, i.e, not providing a benefit.
G.Fast is suppose to sync back up very quickly, so SRA may be even less of a benefit.
BeThere did trial it on ADSL2+ and found nothing but problems, it never went live.
The trouble with introducing things like SRA, G.INP and vectoring after the fact, is that the horse has bolted, by that I mean there is a very broad range of modems out there, all of which have been sold not requiring to support whatever UK/Openreach flavour of these technologies they trial, so it's almost impossible to introduce changes without breaking far more than it fixes.
Regards
Phil
-
I think the only way out for us on ECI cabinets is hoping that OR start pushing G.fast pods on poles and manholes etc. That would solve the inequality of service we're getting as we would be all on the same pods, and if they were pushed closer to our houses we would get a genuine speed boost and be just like everyone else.
I wish OR would comment about the state of ECI cabinets.
I noticed in my area all except 1 are ECI cabinets and they became available in April, May, July, August and October in 2012.
-
I guess the problem with some lines is there are times where there is more noise than a 6db margin can cope with, so it makes no difference if SRA is keeping that margin constant or not, it's going to get knocked out regardless.
That's why the DLM will almost certainly remain on VDSL (FTTC) since SRA is probably not going to suffice alone in cases where the line is unable to sustain after a sudden onset of interference that causes it to drop. After the trial, I would imagine that the DLM would be set the SNR as a "target" and that SRA would just work to get the SNR as low as that target if permitted depending on error ratess. But I could be completely wrong. I only speculate at this point.
G.Fast is suppose to sync back up very quickly, so SRA may be even less of a benefit.
Indeed. Apparently only a few seconds compared to the couple of minutes required for VDSL.
The trouble with introducing things like SRA, G.INP and vectoring after the fact, is that the horse has bolted, by that I mean there is a very broad range of modems out there, all of which have been sold not requiring to support whatever UK/Openreach flavour of these technologies they trial, so it's almost impossible to introduce changes without breaking far more than it fixes.
Very true. I remember the issues that were present with the good ol' BT Home Hub 5A/Plusnet Hub One and G.INP on the upstream. Of course, quite a lot of us know about this ... and these were modem-router combos handed out by some of the largest ISPs (or at least combined, as both are under the BT group)!
I noticed in my area all except 1 are ECI cabinets and they became available in April, May, July, August and October in 2012.
I am guessing that the ECI cabinets were probably part of the commercial rollout? And the one that is not was either part of the BDUK scheme or was put in place some time after the others to boost capacity?
-
Hi
I think the only way out for us on ECI cabinets is hoping that OR start pushing G.fast pods on poles and manholes etc. That would solve the inequality of service we're getting as we would be all on the same pods, and if they were pushed closer to our houses we would get a genuine speed boost and be just like everyone else.
To be fair the speed differences are not that different, the main one is where interleaving is off because G.INP is enabled which can't happen on an ECI cabinet of course, but even then it's not a very noticeable difference. I've been on ECI cabinets at two properties now and always had 80/20 with no interleaving. The biggest decider to speed isn't ECI versus Huawei, it's distance, as always.
There isn't a huge inequality of service, just a small inequality for some connections.
I agree overall that G.Fast should never have been relegated to pods on the PCP, in my opinion any investment into the corroded copper telephone network should always result in some element of removing it, i.e. bringing fibre closer to homes.
G.Folly pods only benefit those people who don't need to benefit from a speed boost.
Regards
Phil
-
Its a domino affect really.
We dont know if ECI is capable of SRA or not, but even if it is, a decision may have been made that the other stuff needs to be working first as a prerequisite and as such no SRA on ECI.
So the falling dominoes would be.
Failed G.INP domino knocks the xDB DLM domino over which then knocks the SRA domino over.
I am curious what will happen to "IP profiles" with SRA, as SRA can change the sync speed every XX amount of seconds. I cannot remember exactly how often it was on my ukonline service but I am pretty sure it was no longer than a minute. On ukonline it didnt matter as they didnt use IP profiles. Also the potential compatibility fallout from this may be interesting as well.
-
So here we are in 2018 saying why didn't Openreach buy V-41's when placing their orders in 2011. Perhaps they may have done if there was such a thing as system level vectoring and V-41's actually available when they started designing their network. It explains why Deutsch Telekom and France also rolled out FTTC with M-41's. I guess you can't plan and order something which isn't yet available. :'(
Do both those telcos still use the M41s and did they use only M41s or switch part way through?
Also SRA when working at its best is absolutely amazing.
My ADSL story was mostly talked about on tbb.
But basically I had an extremely bad line, it was really hideous and I am not exaggerating.
The attenuation was around 50db, and it synced at around 6400kbit on the downstream in optimal daylight conditions on fast path with a 6db SNRM.
In terms of stability there was the following problem.
1 - Been an overhead pole line at 50db attenuation it was affected by the nighttime, the upper half of the ADSL1 tones were decimated, bitswapping would fully disable about 10-15 tones so they remained off for the remainder of the sync, and the bitloading was all shifted to the first 50% of the tones, which under normal conditions should have handled it fine.
2 - Second problem, on working days, so monday-friday every week except bank holidays, there was a sudden large drop in SNRM across about 60% of the useable tones, most of it was focused on the strong lower frequency tones, given after a night most of the higher tones were turned off this was devastating to the line. This would suddenly come on usually between 6.30am and 7.30am, then come and go randomly throughout the day, eventually stopping between 5 and 6pm.
It was a no win situation, if I synced "DURING" one of these events so with absolutely battered tones, I would sync at about 3-4mbit, but then when the first night came the line lost sync, and would sync with full bitloading again on the lower tones, but would be unstable again the following day when the noise burst appeared. DLM obviously took action and the line would end up at a 15dn SNRM "and" interleaving. But still was NOT stable. I eventually moved to AAISP to try and deal with the problem, they improved it by getting me moved to a TI dslam that I believe had a broadcom chipset, but the line was never consistently stable.
Eventually I moved to UKOnline, line no more stable but at least didnt have to deal with DLM. However a guy called Dan appeared on the tbb forums who worked for ukonline and he offered to enable SRA for tbb members. When he activated it on my line it was WOW. I went from resynching on a daily basis with huge periods of high packetloss as well when it failed to resync and just run with massive SES, to sync uptimes of over a YEAR.
It was not 100% flawless, it would go something like this.
6.44am no noise burst.
6.45am sudden noise burst.
X seconds of packet loss and ES
SRA kicks in and sync speed reduced live with no loss of sync.
Service stable no packet loss.
Noise bursts ends, SRA recovers the sync speed NO loss of sync.
BT eventually after a number of years said the issue was below a major road in the town centre on the E side, no chance of it ever been rectified. Given the E side is not used on FTTC the problem doesnt exist now.