Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[digitalnemesis]

Please OR, we need G.fast!

[S.Stephenson]

Unless your less than 400m from the cabinet I wouldn't expect G.Fast till after 2020.

[digitalnemesis]

Unless your less than 400m from the cabinet I wouldn't expect G.Fast till after 2020.

Aren't OR deploying G.fast from the distribution point?

[WWWombat]

How can lines be interleaved but without increased latency? I can see some poeple on mydslwebstats who have Interleaving depth of 16 and 0ms of delay. G.INP enabled of course.

Because interleaving can be configured with many different settings. The softest ones tend to be put in place in parallel with retransmission being activated - where you see depths of  4, 8 or 16. The harshest ones have depths of over 1000.

The delay associated with interleaving comes from it working on a 2-dimensional array of data (with both a width and a depth), so the delay amount is in proporotion to (depth x width) [or (depth x blocksize)] ... the softest settings have a delay of around 0.2ms, while the harshest ones tend to be 8 or 16ms.

[WWWombat]

Unless your less than 400m from the cabinet I wouldn't expect G.Fast till after 2020.

Aren't OR deploying G.fast from the distribution point?

Your guess is as good as anyone's.

Research into G.Fast started life with the expectation it would only be viable at the DP, and has been designed and standardised with that in mind. However, it turns out that the copper has worked better than expected, and longer distances can be used while still allowing for a decent speed upgrade. BT have been asking for improvements to the standards and the chipsets to allow for deployment further away from homes than "just" the DPU; my current belief/expectation/guess is that distances up to approx 300-350m will be the maximum.

That has given OR some relief (they can deploy fewer G.fast nodes, so it will be cheaper), but a conundrum too - where do they choose to put the node now?

So one of the aims of the current G.Fast trials is to find out what speeds happen in reality, and what distances. That will let Openreach plan a deployment.

Until they figure the answers out, we won't know a blessed thing. Once they figure it out, we still won't know - we'll have to guess from watching where they activate nodes.

But...

While they are figuring that answer out, we can speculate a little.

It will almost always be true that they'll have to locate one G.Fast node where the PCP/FTTC node is - to support the premises close to that cabinet. There is already fibre, already power, and already a set of premises. A no-brainer.

Extending that thought process though, we can see that if BT put a G.Fast node at every PCP/FTTC site, and aimed at 300m reach, then they would likely be able to offer service to about 30% of lines in the country - by this: http://postimg.org/image/bp372fcnn/

By coincidence, 30% of lines in the country is BT's target for G.Fast by 2020. ^{Edit: But the coincidence might just be that - and they're planning a different 30% entirely, targetting specific areas.}

If they did that, and only that, then they'd only end up offering the 300Mbps service to people who could already get 80Mbps - politically, lots of people would be up in arms at this choice. And it might not get the best takeup.

It's all speculation, until we see what Openreach end up doing.

[gt94sss2]

I would probably go further than WWWombat in speculating that by 2020 you will see g.fast in the local loop (not just cabinet locations) for areas that are considered economically viable and/or have cable - with little to no g.fast in BDUK areas.

[Chrysalis]

I think its logical they will take the low hanging fruit first, so just deployed at cabinets or next to cabinets.  I have speculated this for months, although I kept been shot down as the trials are not working this way.

I think the people in the worst position is where they over 300m but they only a few per cabinet over 300m, I am by coincidence in this category, my engineer told me only 48 lines out of about 600 are over 300m from the cabinet and i am one of them.  Is it worth BT deploying a node for less than 50 people?

On the plus side crosstalk will ease off on g.fast rollout as people move over and I am happy with vdsl2 speeds.  So its not a major issue for me at this time.

This will also be a factor on any business decision on vectoring for vdsl, because as people move from vdsl2 to g.fast then crosstalk on vdsl2 naturally will go down, there wont be crosstalk from g.fast users with the power cut back.

[digitalnemesis]

At this rate I reckon we'll see 100% FTTP coverage by 2030.

[broadstairs]

Just a quick update on my thread with TT. One of their OCEs ran a line test and said its fine BTOR will not accept this as a fault so basically they are doing nothing. I have pushed for escalation to the CEOs office but I dont really expect to win that one but my thread is getting quite a few views so at least it is causing some interest.

I have only found one other recent thread on the TT community about loss of G.INP but they just commented and no one seemed to follow up on it.

Stuart

[gt94sss2]

On g.fast I am sure I heard BT announce they had chosen their initial chipset vendor (but did not say who) the other day.

If it's Broadcom based, their newer chipsets support both VDSL and g.fast

[ejs]

Openreach have also said they will be adding about 2800 extra cabinets (network re-arrangements), to improve speeds and coverage, started in February 2016, to finish in 2020/21. Presumably some people will be furious that all they got was a closer FTTC cabinet, while others were getting G.fast.

[WWWombat]

I have speculated this for months, although I kept been shot down as the trials are not working this way.

I tend to agree the trial isn't a good indicator yet - Openreach need to use the trial to figure out answers to lots of questions, which needs a variety of deployments that won't always be optimal. The shape of the pilot might tell us more.

I think the people in the worst position is where they over 300m but they only a few per cabinet over 300m, I am by coincidence in this category, my engineer told me only 48 lines out of about 600 are over 300m from the cabinet and i am one of them.  Is it worth BT deploying a node for less than 50 people?

It might be, if they're all the same direction away from the cabinet - which does happen for some of the cabs here in my town.

In circumstances like yours, a single node might not have the capacity for demand from 550 lines ... so splitting at 200m might make sense. In fact, I could see an argument for choosing the range by dividing the total by 2 or 3, rather than rigidly setting range to 300m.

On the plus side crosstalk will ease off on g.fast rollout as people move over and I am happy with vdsl2 speeds.  So its not a major issue for me at this time.

This will also be a factor on any business decision on vectoring for vdsl, because as people move from vdsl2 to g.fast then crosstalk on vdsl2 naturally will go down, there wont be crosstalk from g.fast users with the power cut back.

I do wonder what the demand will be like for Openreach ultrafast products.

That 2014 Ofcom graph told us that about 10% of VM's customers chose the top-tier (120-150Mbps at the time), and another 27% chose the middle tier (100Mbps, or being increased to that level). The 2015 results for VM tell us that 40% were taking the middle or top tiers (100Mbps and 200Mbps, by that time).

If BT find there is only demand from 10% of properties for their ultrafast offering, I don't think we'll see much decrease in crosstalk.

[Chrysalis]

yeah all 48 are bunched together, the coverage from the cabinet is like a U shape rather than O. I am right at the edge of the U, next to the next cabinet.

g.fast will likely be dominated by warez downloaders and people who want to brag I expect.  vdsl2 is fast enough for mainstream.

But then again who knows what marketing can pull off, people can get brainwashed to think they need 100mbps+.

[simoncraddock]

Please OR, we need G.fast!

Do you really think they can pull it off after all the problems with g.inp?
Wishful thinking!

[simoncraddock]

yeah all 48 are bunched together, the coverage from the cabinet is like a U shape rather than O. I am right at the edge of the U, next to the next cabinet.

g.fast will likely be dominated by warez downloaders and people who want to brag I expect.  vdsl2 is fast enough for mainstream.

But then again who knows what marketing can pull off, people can get brainwashed to think they need 100mbps+.

You can have a gigabit service but if the source of your download is capped for QOS its pointless.