Kitz Forum
Announcements => News Articles => Topic started by: niemand on June 29, 2019, 12:40:24 PM
-
[Moderator note: This subject is the result of a merge of two independent topics, one started by CarlT and one started by WalterGMW.]
So, now Andrew at Think Broadband has written about it I suppose I can write a bit too: https://www.thinkbroadband.com/news/8456-openreach-looks-set-to-replace-eci-cabs-with-huawei
Initially replacing of some ECI gack with Huawei SmartAX MA5818 - these are 2RU 4 x half-slot chassis capable of taking 4 x 48 port VDSL cards, 4 x 24 port G.fast cards, capable of VDSL also, or a combination of the two, software permitting.
There also may be replacement with SmartAX MA5616 - these are the small Huawei DSLAMs that have gone from taking 96 VDSL lines to 128 and now 256 with cabinet upgrade.
There'll be an initial pilot to see how things go. Going forward if it's all good it'll be a part of a concerted effort to be rid of this hardware entirely and, with it, remove one of the impediments to removing ECI OLTs/L2S/headends too.
-
This sounds like good news.
I wonder what was the tipping point for OR to consider replacing ECI cabinets? I know people have complained a lot on the forums for quite a few years, so something more recent must have made the case?
I wonder how long the downtime will be when switching over. All the cabinets under the OR rollout in my area are ECI. The only Huawei cabinet was put in a couple of years after under the BDUK scheme. It says properties passed 82. Does that mean 82 people are connected to it, or that its capable of 82 connections?
-
82 properties connected to the PCP that is served by that Huawei cabinet.
Reliabliity issues, cost of attending to faults that wouldn't be an issue with Huawei kit, and probably above all else availability of equipment, the MA5818, that can handle G.fast and VDSL, alongside having a heat output profile that allows it to be installed into an ECI cabinet shell.
As you noted also BDUK cabinets being Huawei means Huawei OLTs are present in formerly all-ECI headends. The ECI OLTs configured in the manner Openreach have them are garbage so replacing them is a bonus too.
I seriously doubt that Openreach could've cared less about complaints on forums or that the ECI kit provided lower sync speeds. That the hardware failed more often, resulted in more customer fault calls and can now be easily replaced likely more relevant.
-
As a sufferer of an ECI cab this is great news. I hope more information will become available however especially on selection criteria to try to get an idea on which or if all cabinets will be replaced.
I am probably on the outskirts (optimistically) of Gfast to my PCP but if the Gfast gubbins ends up in the FTTC cab though I will definitely not be close enough as this is a further 50m away.
I will take Ginp and 3db (and possible vectoring?) over Gfast though :)
-
To think all the times we've been told it will never happen :P
Mind you it hasn't yet, but will be great for those stuck on ECI cabinets if/when it does, especially if the new kit has vectoring enabled :fingers:
I wonder what will happen in situations like ours, all of ours were ECI cab's but lots have big Huawei twins now, I suppose if the capacity of the Huawei is sufficient they could just stop using or even remove the ECI cab, something I did wonder when they started putting in the Huawei twins.
-
Think you misunderstand, Ronski, they aren't going to stop using ECI cabinets they're going to swap out the kit in them. They're going to remove the ECI DSLAMs and replace with Huawei.
Wasn't going to happen but a few things have happened along the way making it an option.
-
CarlT, I didn't misunderstand, not sure why you thought I did, I know they are going to reuse the ECI shell and just replace the innards.
-
Couple of observations by me if they were to swap out the M41s to MA5616s/MA5818s.
Many of the ECI M41s currently support up to 256 lines. The reason they were popular with Openreach is their small footprint and 64 port line cards.
The existing MA5616s are limited to 4x32 ports = 128 lines - Although I believe they are now compatible with the H83BVCMM/H83BVDLE line cards (https://support.huawei.com/hedex/hdx.do?docid=EDOC1000151729&lang=en&idPath=24030884%7C9856746%7C23708799%7C9858846%7C18005) which support 48 ports.
If they did use MA5616, these would not be the same build as the existing MA5616 DSLAM components which have limitations. See my post here on the MA5616 (https://forum.kitz.co.uk/index.php/topic,22764.msg387349.html#msg387349).
There's quite a bit more to making them vectoring enabled than just adding the cage. For starters the existing Central Control Unit Board would need upgrading from a CCUB to a CCUE (more specifically a H831CCUE) to be able to support daughter cards. (to attach a vectoring module to).
The backplane doesn't appear to be able to support this either and _would_ need upgrading from a H831MABA -> H831MABB. You'd also need to swap out the power board with one with higher wattage to be able to add the daughter card (200W -> 400W). There's about only the chassis, fan tray and service board that wouldn't need an upgrade. :( I'm even doubtful about those if they were to have vectoring as the line cards would also without need replacing so not sure how that would affect the system board. Finally the real bummer.. because of heat dissipation if you add a vectoring module, then the MA5616's are restricted to just 96 lines (compared to the current max of 128) which could in itself cause headaches for Openreach unless they housed it in a larger S200.
I don't know much about the MA5818's but a search for the VDSL line cards show that it uses the H83DSDMM Board (https://support.huawei.com/hedex/hdx.do?docid=EDOC1100031381&lang=en&idPath=24030884%7C9856746%7C23708799%7C9858846%7C8967969) which has 48 ports. With only 4 service board slots, this also means a restriction of 192 lines 256 lines :(
Huawei MA5818 video (https://support.huawei.com/enterprise/en/doc/EDOC1100084160)
---
The Magdalene job application (https://search-jobs.magdalene.co.uk/vacancies/2441/night-migration-engineer.html) implies a straight swap out of the ECI kit; installation of both MA5616 and MA5818 DSLAMs in the existing ECI cabs; to MA5600 and MA5800 OLTs.
Final thing to mention is this is a trial and only applies to 200 cabs at present.
----
ETA I was ready to post the above but in-between CarlT advised me there is a newer line card that will support 256 lines. Quickly amended so I can hit the post button before I go out.
[Moderator edited to fix a couple of "adjacent key clipped" typos.]
-
To think all the times we've been told it will never happen
Well, it's not... (yet).
I think many will read this thread and a couple of the comments and assume a mass swap out of all ECI DSLAM's has started or is starting.
It hasn't and it isn't (again, yet).
It was suggested this will initially be tried on around 200 cabinets. That's a tiny proportion of the ECI estate.
I'd be very very surprised if they replaced all 25,000+ ECI DSLAM's with Huawei.
Andrew says in the article
we do not know is what sort of timescale is involved and how many will actually be replaced
This is likely to be very targeted (at least to start with) in circumstances like...
full/nearly full ECI cab, small amount of capacity needed, G.Fast viable area.
Swapping the ECI electronics with a Huawei MA5618 on such a cabinet would add a little capacity, remove the junk ECI kit, and add G.Fast (without the need for a pod).
In some areas OpenReach have ambitions to cover entire exchanges with FTTP
Many ECI cabinets will be removed completely and not replaced as FTTP rolls out.
edit: I'll add...
I think it's great they are making an effort to do something with the inferior ECI kit.
I made a considerable effort to get off my ECI DSLAM and it made a huge improvement to both upstream/downstream.
I would much rather OpenReach rolled out FTTP to make the FTTC cabinets redundant than spend too much time and money upgrading poor kit to what is essentially the same technology.
Fortunately they are doing a bit of both.
The ECI Head-Ends may be a bigger pain than the DSLAM's, and might be part of the reason for this as Carl alluded.
-
Well, it's not... (yet).
I think many will read this thread and a couple of the comments and assume a mass swap out of all ECI DSLAM's has started or is starting.
It hasn't and it isn't (again, yet).
Which is exactly why I said the following
Mind you it hasn't yet, but will be great for those stuck on ECI cabinets if/when it does, especially if the new kit has vectoring enabled :fingers:
-
Well as our ECI cabinet has just had a Huawei twin fitted for expansion looks like we will miss out yet again. BT and there two tier network even though they don't admit it.
-
What a fantastically optimistic view.
They may also, of course, move everyone from the ECI to the Huawei, permitting its switch off.
The Huawei has a capacity of 384 lines with the HD expansion and 512 with the newer expansion.
-
I know sorry but the ECI is crap as we all know. It would be nice if they did and i would be the first one to eat my hat and praise them but think we all know what will happen but will leave it there and hope for a ;D ending
-
Perhaps this article is only make-believe but if indeed Openreach are contemplating ECI replacements I shudder to think of the downtime and costs for what would be very little gain.
Surely true symmetric FTTH as a new-build would be a better option ?
https://www.ispreview.co.uk/index.php/2019/06/openreach-uk-could-replace-eci-fttc-broadband-cabinets.html#comment-206955
For those who have not looked closely the IDC connector blocks differ with partial unused ones to match the different card capacities.
E.g. An ECI card has 64 channels so the last IDC block has 6 unused pairs whereas some Huawei cards were 48 channels so that only wastes two.
I'll repeat my ISP Review comment here too:-
Whilst reconnecting a few fibres is perhaps not a significant engineering task, the concept of stripping out several sets of copper tie cable pairs with different IDC configurations might well be more of a challenge. I suppose one solution might be to add new tie cables to the replacement cabinet ( on a new plinth ??) and then insert standard underground bullet joints – always assuming there’s chamber space.
-
They won't be doing it for the benefit of customers directly. The ECI kit breaks down more often, causes more customer faults and the stuff they connect it back to at the headend sucks without heavy upgrade.
I'm sure it won't be every single cabinet but, in cases like whey there's a Huawei already installed, this programme makes it more likely they will retire the ECI.
When they replace the ECI with Huawei it means the new kit goes back to Huawei equipment in the hubsite so customers have to be reprovisioned. Apart from the need more for tie pairs to the new Huawei it's on many levels an easier job getting shot of the ECI.
Turn it off, save power, save maintenance costs, reduce fault calls.
-
Perhaps this article is only make-believe but if indeed Openreach are contemplating ECI replacements I shudder to think of the downtime and costs for what would be very little gain.
Surely true symmetric FTTH as a new-build would be a better option ?
Walter I read things like that all the time and in part I'm in agreement.
Full fibre can't be the only answer though.
If we all had to wait on FTTP then there's millions who are benefiting from FTTC now who would have had to wait over a decade for it while struggle in their 1-2Mb ADSL.
I'd much rather have my 44Mb FTTC than the 3Mb ADSL I had before.
I would have had years and years to wait for FTTP and that time is right through my child's primary/secondary education when decent home internet is essential.
It could take years to replace all the ECI cabinets with FTTP.
In the meantime they keep breaking down.
ECI kit is less reliable, more fault call outs, parts become more expensive, harder to source.
Someone has done the numbers and it's cheaper over a period to rip it out and replace it with cheaper, reliable, easy to source Huawei kit.
Any down time is overnight only. It's coordinated well in advance with ISP's so any customers with SLA's can have alternative backups put in place.
Many will be over built with FTTP and the DSLAM retired instead of replaced with a Huawei DSLAM.
The cost of maintaining these ECI DSLAM's may simply be too expensive that they can't wait years for FTTP for every single 1.
We're all FTTP fans, but it can't be the only show in town.
-
The Q&A at BT's recent technology briefing (https://forum.kitz.co.uk/index.php/topic,23582.0.html) especially 1:18:33- 1:21:30 (and perhaps 24:30) are among those where Huawei is mentioned - in that while they don't think any Huawei ban will cover DSLAMs, they have thought about what would happen if it did - and that the like having 2 suppliers for everything
-
Thanks for the info Carl
I expect based on my educated guess. Reasons would be similar to following.
Support contract coming to end and ECI's current performance in providing support perhaps been subpar is leading to a possible decision to not renew it and perhaps using unsupported kit is deemed not viable.
The cost of running ECI kit I expect is higher than hauwei probably due to increased levels of faults, because of the lack of g.inp. As well as higher compatibility issues with modems.
Lack of xDB profiles, and general performance leading to hauwei kit been faster which means higher speed estimates and less speed related complaints been made by end users.
I think all these things combined has perhaps finally made it a viable business case to do a swapout.
Ultimately one of the advantages of using 2 suppliers is that if one under performs then it can be replaced with the other, if you not prepared to do that, then it nullifies some of those
benefits.
-
most of the cost will be labour cost.
It would be logical to put in kit that can do more than the hauwei kit installed at start of rollout, so e.g. vdsl 35. But even if its just basic eci -> hauwei I think its misleading to call it "very little gain", the generic advantage hauwei has combined with g.inp, xDB can be significant, in addition g.inp can be difference to having interleave on or off and interleaving is definitely noticeable.
-
I don't think the speed difference or potential improved latency between the 2 vendors will be any part of the reason for this.
Very little gain or significant gain, OpenReach get paid the same for lines connected to either DSLAM vendor.
The lack of G.INP might make the fault level slightly higher on ECI kit but Interleaving would do it's job for the majority.
This will be down to future maintenance costs.
The ECI hardware is considerably more expensive to replace parts, overheats, breaks down more often, and has a higher fault call out rate.
It would be logical to put in kit that can do more than the hauwei kit installed at start of rollout, so e.g. vdsl 35
Agreed, but VDSL2 profile 35b is irrelevant here.
The Huawei MA5618 that will be used to replace some of the ECI DSLAM's can do G.Fast, meaning no additional pod on the PCP would be needed.
The profile 35b ship has sailed. Too late for that.
-
The ECI hardware is considerably more expensive to replace parts, overheats, breaks down more often, and has a higher fault call out rate.
To that list of defects we should also add that those cabinets installed in non-built-up (i.e. rural) areas were found to be too well ventilated and the cold, moisture laden, air caused condensation problems in the card-cage backplane. I seem to recall that heating pads had to be fitted (thus more energy consumed) to stop the condensation build-up.
-
I don't think the speed difference or potential improved latency between the 2 vendors will be any part of the reason for this.
As much as some of us who are on ECI's would like to think it was so, I too suspect it is far more likely to be as you and Carl say, especially if the OLTs aren't too clever either.
After-all if they were just swapping out for the sake of EU performance and rectify some of the M41's shortfalls, then surely it would be far easier to replace them with V41s like Deutsche Telekom and another operator supposedly did with some of their M41s. Therefore there has to be other reasons.
The Huawei has a capacity of 384 lines with the HD expansion and 512 with the newer expansion.
This is a game changer.
Back in 2012 when councils and residents were moaning about street furniture then the ECI's with 256 ports and smaller footprint were the preferable option in quite a few towns.
Since then, there has been the ability to upgrade the MA5603 to 384 lines and now if 512 is on the cards (!) this makes it a whole new ball game.
We are now seeing MA5603 "twins" being installed as the ECI's run out of space, and I suspect there will be more to come. Both councils and the general public are now more accustomed to seeing (or not noticing) cabs so are less likely to kick up a fuss about the size of the MA5603s than they did 7 years ago.
In cases where there are twin cabs or ECI's nearing capacity, surely it would be logical to replace with a MA5603 (512 lines) and decommission the ECI completely. Obviously consideration would be need to be taken into account such as the likelihood and timescale of lines moving over to FTTP in the near future. So there could well be a mixture of options for Openreach to consider - depending upon of course how the trials go.
I think many will read this thread and a couple of the comments and assume a mass swap out of all ECI DSLAM's has started or is starting.
In all fairness, I think that assumption may have come from the linked to TBB article.
It's one of the first things I spotted when seeing the headline and the article itself could lead many readers to believe 25191 cabs could be up for replacement. I'd also read the ISPr article so was fully aware it was just a trial of just 200 - which isn't mentioned in the TBB article.
-
In cases where there are twin cabs or ECI's nearing capacity, surely it would be logical to replace with a MA5603
Certainly not happening yet anyway.
The attached image is a dual ECI site, with the 288 Huawei on the right having just been added.
PCP15 ESDAL.
The original ECI is at the back of the image to the left of the PCP.
It's the only time I've seen 3 cabinets for a single PCP and given recent developments there may not be many more like it.
-
Phase 7b so might've filled the first one and had the second one arrive before Openreach abandoned ECI as a bad job.
-
I don't think the speed difference or potential improved latency between the 2 vendors will be any part of the reason for this.
Very little gain or significant gain, OpenReach get paid the same for lines connected to either DSLAM vendor.
The lack of G.INP might make the fault level slightly higher on ECI kit but Interleaving would do it's job for the majority.
This will be down to future maintenance costs.
The ECI hardware is considerably more expensive to replace parts, overheats, breaks down more often, and has a higher fault call out rate.
Agreed, but VDSL2 profile 35b is irrelevant here.
The Huawei MA5618 that will be used to replace some of the ECI DSLAM's can do G.Fast, meaning no additional pod on the PCP would be needed.
The profile 35b ship has sailed. Too late for that.
You seem to agree and disagree on the same thing, speed complaints, latency complaints, errors due to lack of g.inp all amount to the same thing which is as you said "higher fault call out rate"
You can be under DLM's threshold and still notice visible instability. So basically there will be lines that dont get interleaved but at the same time have things like dropouts and error bursts causing packet loss. Likewise you can get lines looping between fast path and interleaving because the instability isnt consistent. So g.inp can significantly decrease fault reports in my opinion. DLM is not and never will be perfect. It kinda just masks some of the worst instabilities, whilst g.inp is a proper solution, DLM is not its more like a automated band aid. (Remember on ECI the default operating mode is fast path, so interleaving isnt applied on many lines, it requires DLM to identify a line, and then keep it in place. Whilst on hauwei interleaving is default, then g.inp applied after if line fits the stability profile, so by default lines on hauwei will have massively reduced error counts)
Its kind of like the affect SRA had on my old ADSL line, BT's DLM ultimately couldnt fix squat on it, but SRA when implemented transformed the line in terms of usability and stability, and that ultimately halted the fault reports to openreach dead, it went from dozens of callout's a year to zilch once SRA was enabled. g.inp I can imagine has a similar affect on the vdsl product.
So yes openreach wont care about the sync speed as it doesnt directly give them revenue, but they will care about the reduction of callouts that are related to speed. (as well as stability). I suppose I should correct myself actually, there may be some small increases in revenue if xDB can push a line up enough for someone to move from 40 -> 55-> 80 products. The cost of changing the kit is quite low, so even tho that revenue increase wont be huge, it may still be a factor. Also factor in the loss of revenue if someone ultimately cancel's their VDSL and moves to cable/FTTP/4G instead due to performance/stability.
Ronski moved to cable, and I think a factor in his decision was ECI issues.
-
Chrysalis is correct, the poor performance especially on the upstream was a deciding factor in moving to Virgin Media. When I was first connected I got 12Mbps upstream, when I left it was about 6Mbps, would I have stayed if I still had 12 up, possibly.
-
I suspect the the reason for the ECI cabinet change is for;
1. The man power needed to repair and replace parts. There is a shortage of engineers industry-wide, so while OR are also building the FTTP network, adding G.fast pods on to cabinets, fixing G.fast connections, fixing copper line associated faults, the last thing they want to be doing is replacing parts on cabinets.
2. As some have mentioned it might also be something to do with the ECI kit at the exchange. I think OR are looking in to the future and seeing that in order to compete with the likes of CityFibre and others, they need to have the infrastructure to easily be able to increase future speed boosts. Having ECI kit stuck at 330Mbps rate isn't going to cut it when eventually 1Gbps connections slowly become normal, especially on rival networks. So its probably better and more cost effective to sort the future FTTP network and its transition from copper based FTTC now rather than later.
-
My mate Albert who live in Ketley area has suffered VDSL with ECI cabinet (61Meg) for the last 2 years but been replaced to Huawei and now getting (76Meg)
-
My mate Albert who live in Ketley area has suffered VDSL with ECI cabinet (61Meg) for the last 2 years but been replaced to Huawei and now getting (76Meg)
That's pretty much what I would expect based on how my line has deteriorated.
Had 100Mbit on Digital Region as there was zero crosstalk (I was likely the only person on the cabinet due to the terrible uptake), down to 80Mbit on ECI, now chugging along at 66Mbit. I got a second line to make up the difference and that is syncing at 63Mbit. I don't think its unrealistic to suggest I'd still be syncing at or close to 80Mbit if I were on G.INP and vectoring.
-
Openreach statement:
We’re constantly working on ways to upgrade and extend our network but we have no immediate plans to swap out fibre cabinets on any significant scale.
Its said they have about 25,000 ECI cabinets of which they plan to replace around 200.
-
That's usually how trials work.
I wouldn't expect a statement to say any different until the trials done even if their intention is to go ahead with this.
Trying it on 200 cabinets well give them a bunch of info including the time/cost to do each cabinet.
-
Does anyone have any further information? Have the trials started yet?
After suffering 2 weeks of interleaving on my ECI cab before DLM woke up and removed it they can trial using my cab if they like!
Ginp and 3db would give me most of the speed lost due to crosstalk also.
I also believe it is also worth considering that some small value can be attributed to faster sync speeds as it would bring up the average speeds achievable on the BT network. This must be beneficial for advertising, reporting and regulatory reasons.