DSL Checker - G.Fast

DSL checker results for G.Fast look like this:
https://s29.postimg.org/mcr5gbruf/BTW_Checker_Gfast.png

Hope my g.fast line isn't impacted

For those who want to compare speed estimates of FTTC vs G.fast, here are the postcodes being served by that cabinet:

CB1 9BE
CB1 9EH
CB1 9EJ
CB1 9EL
CB1 9EN
CB1 9EP
CB1 9ER
CB1 9ES
CB1 9ET
CB1 9EU
CB1 9EW
CB1 9EX
CB1 9EY
CB1 9EZ
CB1 9HA
CB1 9HB
CB1 9HL
CB1 9HN
CB1 9HP
CB1 9HR
CB1 9HW
CB1 9JG
CB1 9JH
CB1 9JQ

Just a note, that G.fast should only appear *if* the lower clean downstream estimate of G.fast is greater than speed of 80.01Mb/s.

g.fast faster upload than FTTPoD?

Likely the FTTP/OD will change to 50meg up once G.fast goes commercial.

Or G.fast will be offered at a 30meg upload cap

In reply to a post by lee111s:

Likely the FTTP/OD will change to 50meg up once G.fast goes commercial.

I guess it depends whether BT's choice of upstream for FTTP using GPON was related to how much congestion they were willing to accept over the PON.

G.Fast's upstream is not shared in the same way.

In reply to a post by lee111s:

Or G.fast will be offered at a 30meg upload cap

If there was a risk of that, I don't think they'd have come this far with 50Mbps speeds.

Using the first address on that list, no. 1 CB1 9BE gives VDSL (clean) upload speed of 20Mbps and G.fast upload speed of 6.8-10.9 Mbps. Why is it lower?

And who are the first ISP to offer the first customer of the G.fast service?

Is there any DLM involved on the G.Fast service?

The numbers look messed up to me...

DLM - yes
ISPs - I believe BT Retail, TT and Zen were the main ones so far

Also considering on FTTPoD the 30mbit upstream speed is what everyone gets, whilst on g.fast congestion is automatically a lesser isue as various lines will have lower than 30mbit estimates, on kitz someone posted an estimate where the DS speed range was about 95-136 (not impressive) but the eye opener was that on the upstream the same line had 20 20 estimate for VDSL but less for g.fast.

Edited by Chrysalis (Wed 01-Feb-17 19:54:16)

A reminder that BT are currently running on early hardware, and estimates will use data appropriate to that. They need amendment 2 hardware before they can get close to their target ranges, so it is (slightly) pointless looking at actual speed estimates yet. Fun, but pointless.

And for G.Fast, the relative proportion of upstream to downstream can be altered on the fly. If they find the upstream is dropping away faster, they can alter it.

Relative proportion switching on a pod basis or per line basis?

lots of ifs and buts, we can only judge on what is actually happening now.

e.g. VDSL would be much better if only vectoring was deployed, but it is not.

There is no certainty existing equipment would be replaced and BT's history shows they dont like replacing stuff with a short life, instead they like to sweat it.

So what is more likely is the equipment that is deployed will stay as it is and if newer generation equipment becomes available that will be deployed to newer areas. Although granted these are only limited trial areas so may not be subject to normal rollout procedures, they may replace these as there is only very limited amounts deployed.

G.fast as with many other things is benefiting from the increased levels of processing power, and it is conceivable that more of the upgrades are software based these days e.g. improvements in how the vectoring operates and since G.fast had vectoring embedded from day 1 one would have a reasonable expectation that improvements that will a shift from vectoring 48 ports to 96 ports will be software based.

In reply to a post by Chrysalis:

There is no certainty existing equipment would be replaced and BT's history shows they dont like replacing stuff with a short life, instead they like to sweat it.

Examples being?

ECI cabs

And the commercial reason for replacing them all is?

What was the commercial reason for installing them in the first place?

Suspect Huawei couldn't keep up with the original demand during the commercial roll out

Yes, a small thing called a procurement process. ECI were able to meet the needs of BT's procurement for FTTx and they were obviously a similar or lower price than Huawei.

I guess all you ECI critics were vehemently opposed to the contract when it was signed back in 2010...

Any more sensible questions from you today?

In reply to a post by AndyHCZ:

obviously a similar or lower price than Huawei.

Obviously, you get what you pay for. Obviously.

I think they were also de-risking by having multiple suppliers so that if one supplier hit a supply issue they had another to fall back on.

Whatever the reasons back when the contract was signed it doesn't mean us customers on ECI who understand the lower quality service we receive are happy about, especially as we pay the same as someone on a Huawei cabinet, which may even in some cases have vectoring enabled!

I'm lucky in one respect as we've just had a second cabinet installed, which just happens to be a Huawei, so I'll be switching to that once I'm sure it's live.

Incidentally I'm also out of range for gfast from the cabinet, but if/when they deploy further out hopefully I'll be in the right place to benefit.

What lower quality of service do you get?

Edit: I assume we're going to talk about G.INP/retransmission...

When retransmission was applied to ECI DSLAMS, it was found to cause short bursts of intensive noise at certain frequencies. This was fine and didn't impact performance, however a software bug in firmware for modems containing Lantiq/Intel chipsets meant the noise caused the modems to lock up. All parties were notified about this.

Lantiq/Intel developed a firmware update to resolve the issue and ECI developed a firmware upgrade that would detect a modem lockup event and proactively retrain the line.

The modems should be updated with the new firmware (although CP wires-only need to ensure their customers have updated modems). The ECI firmware update and retransmission are due to be released next month (I would guess the same time as the new DLM algorithm is introduced).

Edited by deleted (Thu 02-Feb-17 16:34:37)

How can you guarantee you're gonna get a port on the new cab? If a few people cancel around the same time it's very plausible that you'll get allocated a port on the ECI cab.

he doesnt need to give a reason, you asked for an example and he gave you one.

That is because there is no commercial reason to replace the ECI kit.

The equipment does what it was supposed to do. We would most likely be having exactly the same discussion if another hardware manufacturer provided the equipment instead of ECI.

In reply to a post by AndyHCZ:

That is because there is no commercial reason to replace the ECI kit.

The equipment does what it was supposed to do.

No, it doesn't support upstream G.INP

ECI cabinets will support upstream G.INP.

However (even if they did not), that is still not a commercial decision to spend several hundred million replacing all ECI kit.

In reply to a post by AndyHCZ:

ECI cabinets will support upstream G.INP.

However (even if they did not), that is still not a commercial decision to spend several hundred million replacing all ECI kit.

If they actually start to function like Huawei cabs, offering the same facilities then there is obviously no need to replace them. But when will they?

Edited by deleted (Thu 02-Feb-17 16:57:21)

Their job is to perform a VDSL2 service and they do exactly that.

No one has ever promised you or sold you a service that two different hardware manufacturers will perform identically with the same features.

That expectation is obviously not unrealistic.Why should the features vary by postcode?

which magic mirror did you get that fictional cost from?

Brought at very high bulk e.g. hauwei ma5616 is priced at about 600usd per unit (1000+). ECI units are cheaper.

How many ECI cabinets exist in the UK?

With the figures you have come up with you have claimed the dslams cost more than the fibre itself.

Does the Huawei kit magically make its way into the ECI cabinets?

Around one third of cabinets are ECI and I don't think my numbers would be far off when you consider the equipment, logistics and deployment costs.

When the G.fast node is in operation would it not matter if it was an ECI or Huawei cab as they would be just acting as a pass-through to the node?

Yes that's the one, thanks for the detailed reply although I was aware why it was removed, it actually worked very well on my line which normally has interleaving applied.

It will be interesting to see how my line improves with G.INP and the new lower SNR margins, as it is under performs for the distance from the cab so any improvements will be welcomed.

Here is a question.

What is the business case for replacing these dslams in the early rollout areas? after all g.fast still works right. They are also capable of pumping out 330mbit to 10% of lines connected so there is no marketing justification.

You cannot apply one set of rules to one situation and not to another.

if it would cost BT the best part of a billion pounds to replace ECI dslams they they have some serious inefficiencies in how they work.

In reply to a post by Chrysalis:

which magic mirror did you get that fictional cost from?

Brought at very high bulk e.g. hauwei ma5616 is priced at about 600usd per unit (1000+). ECI units are cheaper.

How many ECI cabinets exist in the UK?

With the figures you have come up with you have claimed the dslams cost more than the fibre itself.

Just out of interest did you cost in the controller cards and line cards or are you just talking an empty backplane?

There are no guarantee's but there are things that can be done to improve your chances. Apparently once the new cab goes live then all new orders will be provisioned on that. But if I used my existing line, cancelled FTTC and then re-ordered on my existing line, there's a very good chance it would still be connected to the ECI cab. Ordering an ADSL service in the interim would solve this problem, or as I intend to do simply order a new line. There is one guy on Kitz who's line should be connected tomorrow, and we'll get to see if he's managed to get onto a Huawei cab.

There's quite a bit of info in this thread on Kitz

If the G.INP fix works, and and the new SNR margins work just as well on ECI then I may not change, I'll see how things go.

Won't matter as the FTTC is just providing power for the g.fast box and access to the fibre already within

I think you've rather exaggerated what he said, apparently there is 77,033 cabs live, so a third would be 25,677.

If I guessed at £5k each that would be £128,385,000, but I'm guessing it would cost more as you couldn't just replace a cab as that would cause loss of service for probably many days whilst things were changed over so they may have to install a new cab with all the addtional work as well.

Alas it's never going to happen, and if G.INP and the new SNR margins work then there's no more problem (hopefully) and hopefully we'll find out pretty soon.

I guess additional Huawei headends would be required as well

Nothing to do with it.

Really?

I'm not quite sure how a few hundred million pounds equals best part of a billion pounds.

The original BT contract with ECI back in 2010 was worth £528m. This number is likely to have increased with BDUK because the original contract would have been for BT's commercial deployment.

What you are forgetting about the cost of swapping all ECI kit with Huawei is that it's not just the hardware, there are many other costs as well, such as the capital labour element (engineers doing the cabinet/exchange swap work, the cost of TSO staff doing the reconfiguration). Then there is the matter of compensation for ISPs for the downtime (say a 256 cabinet is full and BT pay a month's line rental, that alone would cost them a few thousand).

I don't see why there is any talk of replacement of FTTC cabinets.

G.fast pods will be available for ECI cabinets.

thanks for the cab count, so about 15 million then.

will get back to you, have asked the sales rep

yeah but these cards dont have the same value now as they are now older generation spec.

Also we dont know if that contract was fully fulfilled.

The equipment in the cabs takes an optical feed from the Openreach equipment in the exchange. From that it produces a VDSL2 signal which it merges with the phone signal coming down the copper from the PCP.

And vice versa.

The Huawei and ECI equipment do exactly the same job. What is supplied to them, and they return to the exchange, is identical. What they send to the user and receive back was supposed to be identical. However we have found that the ECI kit in several respects cannot handle the Openreach product developments since their original selection and rollout. It cannot add those improvements to the customer's signal.

So yes, "really" .

I think in practice the 2 won't communicate with other.

Which two won't communicate?

Are you saying each type of cabinet needs it's own manufacturer's GEA headend equipment?

Cabinet and headend if trying to mix vendors

Bahhhh like others I waited years for FTTC/P and get the ECI cab. Around the corner there are Huawei cabs all fitted within about 8 months of each other.
Why?

Will there be G.fast pods for AIO cabinets or can an AIO have a mix of VDSL2 and G.fast cards internally?

Is that a yes? Or are you saying the headends are the same but if two types of cabinet are serviced by the exchange there needs to be a two identical headends, one for each type of cabinet?

I could believe that, but wonder how it comes about.

Does it get even more complicated on exchanges with both FTTC and FTTP, and worse still when postcode areas mix, and even worse when individual lines can get either?

On the face of it don't see why not. There is space inside for additional ties to the pod.
However, as most if not all of the AIO have been either BDUK or pravivately funded, it seems unlikely they will be upgraded to g.fast any time soon

Edited by witchunt (Fri 03-Feb-17 10:24:06)

The guy on Kitz has now moved to the Huawei cab, he's on exactly the same line and has seen almost 4 mb increase in upstream.

http://forum.kitz.co.uk/index.php/topic,18678.msg342...

In reply to a post by AndyHCZ:

ECI cabinets will support upstream G.INP.

However (even if they did not), that is still not a commercial decision to spend several hundred million replacing all ECI kit.

are you sure about that? can you produce anything to say that's the case? ECI themselves say it doesn't. The white OpenReach ECI modems support upstream G.INP, but the M41 DSLAM doesn't.

I totally agree with the 2nd part of your post. There's no commercial case for swapping out ECI cabinets. They generate the same revenue as Huawei cabinets and 99% of EU's won't know there's even a difference.

Edited by j0hn83 (Sat 04-Feb-17 08:10:09)

From internal Openreach presentation.

Where have ECI said VDSL DSLAMs do not support upstream retransmission?

In an email from Ian Lawrence, that I would be happy to share privately. OpenReach have confirmed this a number of times also.

In reply to a post by WWWombat:

In reply to a post by lee111s:

Likely the FTTP/OD will change to 50meg up once G.fast goes commercial.

I guess it depends whether BT's choice of upstream for FTTP using GPON was related to how much congestion they were willing to accept over the PON.

G.Fast's upstream is not shared in the same way.

In reply to a post by lee111s:

Or G.fast will be offered at a 30meg upload cap

If there was a risk of that, I don't think they'd have come this far with 50Mbps speeds.

http://www.ispreview.co.uk/index.php/2017/02/bt-upda...

"However it will come as no surprise to find that BT also intends to introduce two additional FTTP profiles in order to match their G.fast service: 160Mbps (30Mbps) and 330Mbps (50Mbps). BTWholesale are currently expecting to start offering these from May 2017."

Edited by deleted (Mon 06-Feb-17 09:42:33)

In reply to a post by R0NSKI:

Whatever the reasons back when the contract was signed it doesn't mean us customers on ECI who understand the lower quality service we receive are happy about, especially as we pay the same as someone on a Huawei cabinet, which may even in some cases have vectoring enabled!

I'm lucky in one respect as we've just had a second cabinet installed, which just happens to be a Huawei, so I'll be switching to that once I'm sure it's live.

Incidentally I'm also out of range for gfast from the cabinet, but if/when they deploy further out hopefully I'll be in the right place to benefit.

How do you propose to switch to the Huawei cabinet then? You'd have to be ablsoljtely sure that the ECI was full if you plan on ordering a second line!

The cabinet has been switching between available and waiting list for someone now, it's now constantly showing a waiting list. If it goes the same as John's on Kitz it will soon show an available date, which is the date they start accepting orders on the new cab. John has now switched to a Huawei cab.

Like most things in life there is a risk., and I accept that.

Hi Andy, perhaps you could shed some light on this ISP Review article which states ECI will not be getting the new 3db profiles?

http://www.ispreview.co.uk/index.php/2017/02/bt-upda...

Thanks.