Kitz Forum
Announcements => News Articles => Topic started by: Bowdon on August 09, 2018, 04:04:57 PM
-
https://www.ispreview.co.uk/index.php/2018/08/openreach-to-scale-back-rollout-of-g-fast-broadband-focus-on-fttp.html (https://www.ispreview.co.uk/index.php/2018/08/openreach-to-scale-back-rollout-of-g-fast-broadband-focus-on-fttp.html)
Industry sources have informed ISPreview.co.uk that Openreach (BT) are to revise down their UK rollout plan for 330Mbps capable hybrid fibre G.fast broadband ISP technology from 10 million to nearly 6 million premises. The aim is to refocus on their new Gigabit capable Fibre-to-the-Premises (FTTP) deployment.
Under the previous plan Openreach had aimed to push G.fast coverage to 10 million UK premises by 2020, while their “Fibre First” deployment would work to extend “full fibre” FTTP to 3 million premises within the same time period (plus they’re still working on an aspiration to reach 10 million premises with FTTP by around 2025 at a cost of c.£3-5bn).
At the time the operator made it clear that FTTP and G.fast would be rolled out into different areas, although in practice this could be quite complicated to achieve (the limited coverage and variation in G.fast performance isn’t always easy to pin down). On top of that the Government, Ofcom and many others have been continually pushing FTTP for the future.
Nevertheless BT Group recently reported that G.fast has so far covered 1,117,000 premises, while their FTTP rollout had reached 631,000 premises, but the latter seemed to be ramping up much more quickly. Meanwhile the G.fast deployment appeared to be going much more slowly than it should have been at this stage.
An Openreach Spokesperson told ISPreview.co.uk:
“Full fibre’s our priority and we accelerated our investment plans to reflect that. Our engineers are already building FTTP to 10,000 premises each week and we’re on track to reach three million homes and businesses by the end of 2020. Our ambition is to reach 10 million by the mid-2020s.
We’re keen to make ultrafast broadband available to as many people as possible, as quickly as possible. Gfast allows us to do that alongside our big full fibre build with little disruption to local communities, so we plan to upgrade more than five and half million premises using that technology.”
-
Well this wasn't entirely unexpected imho.
Something must have happened for all those G.fast pods attached to cabinets up and down the country, all showing available and suddenly all disappearing from the checker.
-
This is good news. Some logic has hit Openreach? :D
-
I've never been a real advocate for G.fast. The deployment model that they decided to use was detrimental to having ultrafast coverage other than for perhaps up to ~60% premises. Of course, this figure would be subject to a reasonable amount of variation, though I imagine they were only ever going to deploy G.fast to higher density areas (where a good amount of premises should be able to receive ultrafast speeds of 100 Mbps or more).
VDSL utilising up to 35 MHz (whether that would be SuperVector or VPlus, though the latter probably less so as it is nothing something compatible with any existing hardware) should have been the intermediate plan since it should allow for subscribers at up to twice the distance from the cabinet compared to G.fast obtain 100 Mbps. I haven't managed to take the time to understand every aspect about Huawei's SuperVector technology, but what I do know is:
- It is compatible with MA5603T
- It can co-exist with existing 17a profile
- It can provide 100 Mbps downstream up to around 700m
However, it is not clear (perhaps I have missed something/I do not understand entirely):
- What the maximum subscriber density is (in terms of ports)
- What sort of cabinet hardware upgrades would be necessary
- Whether changes would need to be made to the backhaul to accomodate increased bandwidth
Anyway, this is coming from someone who has no experience with DSLAMs or network engineering, so a lot of this could be wrong. Perhaps it is completely uneconomical to consider, and perhaps because it has never really been publicly noted that there has been thought put into the potentiality of rolling it out then maybe no testing was ever done, and Openreach would need to spend perhaps the best part of a year testing it alongside existing hardware, etc. (which in that time they may as well just commit to G.fast and FTTP rollouts where appropriate).
In reference to SuperVector, someone could do some reading here (http://support.huawei.com/enterprise/en/doc/EDOC1000137144 (http://support.huawei.com/enterprise/en/doc/EDOC1000137144)) if they were interested.
The above is not to say that I wouldn't like to have G.fast, because that's not true. I would benefit greatly from having it since I have a somewhat short loop. There is at least one cabinet which has a G.fast pod on, but it is not mine (which needs a re-shell) and this latest news may even mean that I won't get it despite showing as planned on the checker previously (before planned G.fast records were mass-removed). Though I shall cross my fingers and hope that it will come anyway.
-
I've said from when they picked G.Fast that they should have gone with VDSL2 profile 35b.
It would have increased speed and reach on FTTC for everyone, not just those very close to the cabinet.
I suspect the ECI estate played a huge part in that decision though.
The existing Huawei cabinets could have Vectoring deployed and be made to work with a Profile 35b rollout (even with other vendors) but the existing ECI DSLAM's would need replacing for this to happen.
-
Yes I agree that perhaps ECI DSLAMs were part of the reason why they went down this path. There is a large ECI DSLAM footprint, especially from some of the initial commercial rollouts (around the time when ECI was favourable). This would lead to essentially what is known as a postcode lottery on whether you could exceed 100 Mbps on FTTC which I imagine they would want to avoid, even though a decent portion of people could benefit from it (probably more people than can benefit from G.fast to get ultrafast speeds with less effort and spending). But I guess they can't compete with the likes of Virgin Media and alt-nets in commercial rollout areas that have ECI without the ability to provide anything faster than 80 Mbps.
In reference to compatibility with SuperVectoring, I can see that the MA5616 is listed in the Huawei docs, but not the MA5616T (the model mentioned on the Kitz site https://kitz.co.uk/adsl/fttc-cabinets.htm (https://kitz.co.uk/adsl/fttc-cabinets.htm)). Perhaps there is an error here? I can't see the MA5616T in the Huawei support at all under MDU (http://support.huawei.com/enterprise/en/category/access-network-pid-1482605784817 (http://support.huawei.com/enterprise/en/category/access-network-pid-1482605784817)), unless perhaps the "T" version is a minor revision. I don't know.
-
Is installing new G.fast pods really that much less work than replacing ECI DSLAMs? Would the problem be that a concrete base for an ECI might not be big enough for one of the larger Huawei cabinets (thinking about more rural cabinets in verges, no so much ones on pavements)? I thought the biggest expense was getting the fibre optics and power to the FTTC cabinet, not so much the box of electronics on top.
I think profile 35b would require replacing the line cards in the Huawei cabinets, and of course fitting a new vectoring engine. Somehow replacing almost the entire contents of a Huawei cabinet is perfectly fine, but replacing an ECI is a no go because it's an ECI and ECI are rubbish, indestructible, and Openreach have lost all the keys to get into them? :P
A bit like not getting a new computer CPU because you'd need a new motherboard, or just getting a new CPU and motherboard. Or building a whole new computer and sitting it next to your existing one.
-
Or building a whole new computer and sitting it next to your existing one.
Yep, thats what I've done ;D But the new one is my main PC now.
I think a quite a lot of the ECI cab's round here now have Huawei twins.
-
What I implied that was upgrading hardware in existing Huawei cabinets is probably cheaper and going to cover more premises with ultrafast than G.fast. I could be wrong, since I do not know what these upgrades will cost. I wasn't intending to imply replacing ECI cabinets with Huawei would be cheaper than G.fast, but the point still stands that SuperVectoring would provide more premises with ultrafast than G.fast if it were rolled out to cabinets.
-
So what did the ECI cabinets have to do with the decision to go for G.fast then? Was the logic something like, it's a bit more work to replace the entire ECI cabinet with some other cabinet, so lets install new G.fast equipment instead, which would possibly be just as much work if not more?
Also, does anyone really care about just barely scraping over the 100Mb ultrafast threshold?
-
Openreach should have rolled out 100% Huawei cabinets and bring in SuperVectoring with VDSL2 profile 35b from 17a in the first place rather than G.Fast cos it can get 300/50 within 300m.
https://en.avm.de/guide/high-speed-and-range/35b-supervectoring-explained/
[attachment deleted by admin]
[attachment deleted by admin]
-
I am amazed. But then I am also ignorant. What an embarrassing cockup about-turn. But sanity arrives belatedly and this is very very good news. If they ever were to put a green cab of some sort in Heasta it would be in the centre of the village and so I would be too far away from it, as the village is a long strip about 1200m long or something, and I am the northernmost house, so if it were still using copper then I would be relative knackered compared to others and especially G.fast would be a joke.
Btw, would that ever happen in this future ? If BT were to put an entirely new node into a village that was all-EO lines only before, would it in future be FTTP since they need to do so much replacement anyway? Or is there still going to be a danger of future FTTC nodes in such situations?
Perhaps I should beg some politician or other to do the right thing? Or perhaps forget the whole thing and suggest they spend the money on something important such as the NHS, old folks or whatever.
-
:silly:
https://www.ispreview.co.uk/index.php/2018/08/openreach-to-scale-back-rollout-of-g-fast-broadband-focus-on-fttp.html
Been better to have gone for 35 profile than G Fast.
-
ejs, it was just speculation on my part. I would take anything I say in reference to economic viability with a pinch of salt since I do not know the true costs nor reasoning behind the choice of a large company. I imagine G.fast hardware alone is expensive, but surely the work associated with replacing a cabinet and its hardware would take more man hours, scheduled works (pathways/near roads) and more materials? And you've got distruption for the EUs from replacing the cabinet. But I imagine existing nodes may compatible with new VDSL hardware, so that's one less issue (but a node upgrade may be necessary to prevent severe contention). But what should be asked is that perhaps due to the way telecommunications work in the UK, Openreach probably want to be careful not to create too many diverse mainstream products that ISPs need invest in?
All this ignores the technologicial limitation of 96 ports perhaps being the maximum for G.fast in the foreseeable future (as I imagine vectoring is a huge challenge with such bandwidth at a high density). That's bearing in mind that the technology was intended to be used at distribution points since each line card supports 24 ports and I believe the current maximum is 2 line cards (48 ports), though there are not many cases where DPs exceed those amounts (but sadly, this is not the deployment case). I do not know what the density for SuperVectoring is, so I cannot comment about that (but I was under the impression it was less than standard VDSL2). But putting future G.fast against current maximum FTTC densities per DSLAM (96 vs 384 possible subscribers) and if we say there are exactly 384 lines under a typical distance scenario then even though up to ~240 should be short enough for G.fast (100 Mbps minimum) then only ~40% of those close enough (25% of total lines) could benefit from it at the same time (but of course this assumes demand).
It's a good question to ask whether anybody cares about just scraping 100 Mbps (which is the current ultrafast threshold). It's a numbers game. Thresholds and limitations deciding what technologies can be labelled as. I just understand that the connection will go as fast as it can based on many variables. But the threshold impacts what the technology can be sold as and who it is available to. Just like how BT couldn't sell their superfast products to those below the 24 Mbps threshold - it had to be a seperate product, even though it was still FTTC.
-
So what did the ECI cabinets have to do with the decision to go for G.fast then?
The fact they would need removing/replacing for a Profile 35b rollout, probably a big part.
Also, does anyone really care about just barely scraping over the 100Mb ultrafast threshold?
Barely scraping 100Mb? I think you need to look up profile 35b.
-
Openreach should have rolled out 100% Huawei cabinets
It is so easy to say that now with the benefit of hindsight. :( You do know that ECI were the first manufacturers to make a DSLAM capable of system based vectoring?
Back in 2012 when Openreach were already busy installing the M41's (which had been ordered a year or so previously) who had actually heard of system based vectoring?
G.993.5 was a relatively new invention and previously only worked at line card level. It was complex and wasted ports on the DSLAMS - so vectoring wasn't really a viable option until ECI manufactured the worlds very first Vectored DSLAMs in 2012.
Openreach were not the only SP caught out by the advances in a new technology both Germany and France also ordered M41s - which at the time were actually the preferred DSLAM due to its small footprint and capable of serving more lines
There's an article here (https://broadbandtrends.wordpress.com/category/vdsl2-vectoring/) about Australias's NBN rollout of FTTC (which was after Openreach's) which states
at the time of the initial planning for NBN – technologies such as DOCSIS 3.1, VDSL2 Vectoring and G.fast did not exist (except perhaps on paper and in the lab).
So tell me how in 2010/2011 when Openreach were ordering hundreds of M41s they were supposed to have a crystal ball that none of the other SPs had either :/
--------
The problem with the M41's is there is insufficient space on the backplane to install a vectoring unit.
Would a simpler solution be to swap out the M41s with V41s? Same footprint, same linecards. I believe thats what Dueche Telecom did with some of their M41s.
-
Well this wasn't entirely unexpected imho.
Something must have happened for all those G.fast pods attached to cabinets up and down the country, all showing available and suddenly all disappearing from the checker.
Still not working,well mine isn't, was installed July/August last year. :D
-
Bowdon beat you to it.
https://forum.kitz.co.uk/index.php/topic,22178.0/topicseen.html
-
The fact they would need removing/replacing for a Profile 35b rollout, probably a big part.
Barely scraping 100Mb? I think you need to look up profile 35b.
Thanks for that image J0hn. Whilst you don't get the massive speed improvements at 100m, like you say Vplus would be more beneficial for far more premises.
(https://forum.kitz.co.uk/index.php?action=dlattach;topic=22178.0;attach=25853;image)
-
VPLUS is not a Huawei technology (SuperVector is) but the range and performance is similar and both can co-exist with 17a. SuperVector is said to be effective until around 700 meters to supply 100 Mbps, but beyond that point is is typically not able to supply 100 Mbps downstream. Going by distance statistics from thinkbroadband (https://www.thinkbroadband.com/how-broadband-works (https://www.thinkbroadband.com/how-broadband-works)), SuperVector could supply 100+ Mbps to almost 90% of premises (probably anywhere between 70-90% depending on line conditions, material and gauge). 200+ Mbps should be possible within 500 meters.
-
I deliberately didn't mention Vplus, or Supervector for that matter.
I just couldn't find a graph that says profile 35b so the VPlus 1 was good enough.
VPlus and Supervector are just marketing dribble though.
They both use profile 35b, and are both compatible with 17a.
Wether they work alongside each other I have no idea but it's just marketing dribble either way.
I've never seen vectoring across multiple DSLAM's so between vendors it's highly unlikely
-
I deliberately didn't mention Vplus, or Supervector for that matter.
I just couldn't find a graph that says profile 35b so the VPlus 1 was good enough.
I absolutely understood that and appreciate that the similarities between them do not detract from the effectiveness of vectored 35b.
-
The fact they would need removing/replacing for a Profile 35b rollout, probably a big part.
Which would have been more work?
1. Upgrade the Huawei cabinets and replace the ECI cabinets.
2. Install G.fast pods everywhere.
Barely scraping 100Mb? I think you need to look up profile 35b.
This was referring to the wonders of getting 100Mb out at 700m, not for the people right next to their cabinets.
Of course, if you take a graph from a G.fast vendor, and got rid of the VDSL2 so that G.fast could use the whole frequency range upwards of 2.2 MHz, things look a little different:
-
Of course they look different but that's not how it's being deployed.
I wasn't arguing 1 was more or less work than the other.
I was pointing out the benefits.
This was referring to the wonders of getting 100Mb out at 700m, not for the people right next to their cabinets.
I'm sure everyone at 700m would be happier with 35b than G.Fast though. I don't think you would hear any of them complaining at 100Mb.
The point of G.Fast was supposed to be that it was taking fibre deeper in to the network. I was all for that.
A G.Farce deployment from the cabinet is useless, idiotic, and I'm not surprised in the slightest they are back tracking on those plans already.
-
but surely the work associated with replacing a cabinet and its hardware would take more man hours, scheduled works (pathways/near roads) and more materials? And you've got distruption for the EUs from replacing the cabinet.
And what did fitting all those G.fast pods involve?
You'd still have some disruption for the EUs even if you're only replacing all the line cards and vectoring engine in a cabinet. I thought someone said some of the older Huawei cabinets would actually need more extensive upgrading to support vectoring.
-
I don't think you would hear any of them complaining at 100Mb.
Now this I am completely sure cannot be correct. They'd be some people complaining. There are enough people who complain about getting a couple of Mb short of the full 80Mb! :P
-
Rolling out 35b would have been no quicker than rolling out g.fast - rollouts in places like Germany are years behind schedule.
g.fast actually makes more sense to deploy and is easier to rollout deeper into the network (as was the original plan). Plus the fact, that its still being developed (longer range, faster speeds, more frequencies) - thats why many more telco's have opted for it instead of 35b.
It won't have anything to do with the fact that some FTTC DSLAMS are ECI and others are not - and replacing the ECI ones is more complex then just replacing the DSLAMS - other kit would need to be replaced too. Finally, there are very good reasons why BT always chooses more than one supplier for its kit.
-
Now this I am completely sure cannot be correct. They'd be some people complaining. There are enough people who complain about getting a couple of Mb short of the full 80Mb! :P
There's always one. ::)
-
g.fast actually makes more sense to deploy and is easier to rollout deeper into the network (as was the original plan).
That is the key line.
If it had been rolled out as originally planned then I think the buy rate would be higher, and the cable maintance would have been lower. At the moment OR have the worst of both worlds.
G.fast should have been used in conjunction with full fibre as was originally planned and then it would have been a genuine step forward. At the moment it looks to me like it is more a tick box exercise.
Apart from saying X amount of people can get 100Mb speed doesn't mean X amount of people will actually buy it.
-
G.fast was always intended to be rolled out like so - like from a DPs to cover premises that were typically no further than a couple hundred meters away at most. Or perhaps even in FTTB scenarios where apartments are involved.
-
I'm sure everyone at 700m would be happier with 35b than G.Fast though. I don't think you would hear any of them complaining at 100Mb.
I bet thousand of customers within 700m are happy to take 100Mbps from Profile 35b if all cabinets are Huawei 100% with supervectoring rather than G.Fast
-
That is the key line.
If it had been rolled out as originally planned then I think the buy rate would be higher, and the cable maintance would have been lower. At the moment OR have the worst of both worlds.
G.Fast is being rolled out as was originally intended - it was always going to be most economic and easiest for Openreach to start the rollout at the PCP's and then deploy it deeper in the network. This would have been the case even if they hadn't managed to increase the range of G.Fast - its just that certain people don't have the patience..
Ironically, the increased focus on FTTP and the scaling back of g.fast will mean that fewer people will be covered by 100mb+ speeds in the next few years then was previously planned.
-
G.Fast is being rolled out as was originally intended - it was always going to be most economic and easiest for Openreach to start the rollout at the PCP's and then deploy it deeper in the network. This would have been the case even if they hadn't managed to increase the range of G.Fast - its just that certain people don't have the patience..
Unless I have missed a document at the announcement of G.fast I was under the impression the original idea was to put G.fast pods on nearby poles and underground, which would have pushed deeper in the network as a stepping stone for FTTP. Then the idea was halted because of power issues i.e. who would power them, the cabinet or the customer. So the pole/underground idea was dropped and they moved them back to the PCP.