News of trial Openreach 1.2 and 1.8Gbps FTTP speed tiers

Interesting trial:
https://www.ispreview.co.uk/index.php/2022/09/openre...

New Nokia & Adtran ONTs with 2.5Gbps interface:
https://www.ispreview.co.uk/index.php/2022/09/pictur...

Still on GPON rather than XGS-PON, just squeezing it more.

Personally, I think them trying to increase FTTP speeds up to 1.8Gbps on GPON is really not good, lets hope the pilot fails and they instead move anyone who wants above 900Mbps onto XGS-PON.

The sixty four million dollar question is when Openreach will begin migration to XGS-PON (or indeed subsequent updates).

But for now it’s GPON only for them and they could squeeze more out of it as CityFibre do with their GPON network, running it symmetrically for example.

Most PONs actually aren’t as heavily subscribed as what you may think, so in fact have quite a bit of capacity. @XGS_Is_On has previously commented on this fact.

In reply to a post by Pheasant:

The sixty four million dollar question is when Openreach will begin migration to XGS-PON (or indeed subsequent updates).
But for now it’s GPON only for them and they could squeeze more out of it as CityFibre do with their GPON network, running it symmetrically for example.

I thought CityFibre did announce they are upgrading all the installed to XGS-PON already, which makes sense. Hopefully OR will get that approved by the "financial controllers" and move to XGS-PON deployment in the near future!

Most PONs actually aren’t as heavily subscribed as what you may think, so in fact have quite a bit of capacity. @XGS_Is_On has previously commented on this fact.

I wonder if this is because at the moment FTTP is sold for "increased speed", at a higher cost ? I know many people whom are very happy with 40-50 Mbps, but would like increased reliability for the same price. Then takeup could increase, but at those lower speeds plenty of capacity.

And I selfishly hope they don’t spend a penny on XGS-PON until those of us missed by the current fibre rollout get fibre 😢

In reply to a post by andynormancx:

And I selfishly hope they don’t spend a penny on XGS-PON until those of us missed by the current fibre rollout get fibre 😢

Well I hope when they deploy in my town that they use modern XGS-PON hardware instead of 5 year old GPON hardware, but yes don't add any delays due to XGS. We have an alt-net deploying across town (other side of motorway) but no sign of OR.

One interesting point is that GPON is split 30 ways, and XGS-PON can be split 128 ways. So for downstream the actual bandwidth available per user is similar. I'd hope that openreach and the altnets all have monitoring in place, and have a good idea as to the spare capacity.

I do also wonder how much the contention on the PON impacts things vs the contention for backhaul and the internet in general.

Also domestically it must be getting more and more difficult to make full use of the higher speeds for a prolonged periods of time. I have a 350MB service but even with a mesh WiFi system I struggle to get above 100MB in all but one room. I expect most people wouldn't notice speed drops due to contention up to a point. This could of course all change when a killer app is found that needs high speed

It wouldn't surprise me if most of the new installs of OLTs weren't XGS-PON ready, or easily upgraded. Its a shame that the ONTs being installed are GPON only

In reply to a post by mwarby:

This could of course all change when a killer app is found that needs high speed

The people whom seem to be wanting Gigabit are those whom download those insanely large games using either PS5/XBox or a PC and they're all using Ethernet connections.

Show how out of touch I am with gaming lol, I thought it was still mostly optical media. Still hopefully there would be enough randomness to minimize the impact. A bigger problem could be P2P file sharing as that could have high sustained usage, but I think its less of a problem now than it was

Just found the announcement from late July. Looks like they’ve done the pilot in York and will be rolling out new subscribers onto XGS-PON as standard from April next year. Looks like Nokia will begin their XGS-PON kit deliveries to them in the next quarter.

Unless I missed it, no specific mention of how/when they will upgrade all their existing subscribers.

Keeping up with the AltNets!

I'd guess the OLT will support GPON and XGS-PON at the same time and new installs will get XGS-PON ONT. Older installs probably getting updated when there's a fault, or if customer asks, eventually they'll have a program to upgrade the remaining users.

Of course probably within the decade XGS-PON will get swapped out for 50G or similar

In reply to a post by jchamier:

In reply to a post by andynormancx:

And I selfishly hope they don’t spend a penny on XGS-PON until those of us missed by the current fibre rollout get fibre 😢

Well I hope when they deploy in my town that they use modern XGS-PON hardware instead of 5 year old GPON hardware, but yes don't add any delays due to XGS. We have an alt-net deploying across town (other side of motorway) but no sign of OR.

I know what you mean! I've got friends and family still on ADSL and another getting 3Mbs on FTTC. The future is not evenly distributed.

Sure but I can’t see Openreach altering their split ratio - it is what it is - so will stay as 32-way with up to 30 subscribers.

When they decide to upgrade they already have the necessary WDMs in place to allow a seamless shift so that GPON and XGS-PON can coexist on the same PON. In future the can step from one technology to the next in the same manner. So XGS-PON can coexist with 25 or 50GPON etc etc.

Ooops. Just replied to your earlier post, pretty much along the same lines.

It depends how Openreach (or any other operator) want to do it. The more modern OLT’s they have like the Huawei, Nokia and Adtran support mixing and matching of cards on a backplane chassis. The cards support either GPON or XGS-PON with the appropriate pluggable optics installed in the ports.

They may opt to install whole new chassis as part of an upgrade path. It’s very likely Huawei will be left as is and not upgraded due to now established supplier bans and limits.

I was meaning just because an altnet such as city fibre or netomnia (who are hopefully going to cover my property soon) uses XGS-PON it doesn't mean its massively better for the end user.

I expect in time openreach will move to XGS-PON and those currently on XGS-PON will upgrade to 25/50G PON or reduce those split ratio as required to support higher end user speeds

Well one is genuinely symmetric and the other can be made to appear symmetric. That to me is a pretty big way to being massively better 😀

If you don’t alter the split then your effective bandwidth per subscriber is an order of magnitude better as you make your way up the generations.

Edit: all else being equal, onward backhaul connectivity needs to increase in similar steps for this to be of any benefit.

Edited by Pheasant (Sat 17-Sep-22 18:14:12)

Our FTTC is steadily getting worse. Was 30/5 when we moved in 9 years ago, neighbours report 11/1 now (we’re on Starlink+LTE for now).

That's quite a degradation. Presumably due to crosstalk as new subscribers sign up.

Partly, but not entirely.

We are a group of six houses, on the end of a 1,600 metre section of buried cable. OR admit the section has many fault and over the years they’ve moved lines between different pairs to try and find the best ones.

When it was just me that had FTTC it was 30/5. When the neighbours joined it was 25/5 but now, with two fewer lines using FTTC (I had two FTTC lines) it is under half that speed.

And we got no broadband county funding, because as far as OR were concerned our lines were capable of 30+ 🤣

In reply to a post by mwarby:

One interesting point is that GPON is split 30 ways, and XGS-PON can be split 128 ways. So for downstream the actual bandwidth available per user is similar. I'd hope that openreach and the altnets all have monitoring in place, and have a good idea as to the spare capacity.

I do also wonder how much the contention on the PON impacts things vs the contention for backhaul and the internet in general.

https://www.ispreview.co.uk/index.php/2022/09/openre... might be an interesting read on this.

In reply to a post by mwarby:

One interesting point is that GPON is split 30 ways, and XGS-PON can be split 128 ways.

Both can be split 128:1
Many networks in Europe do GPON at 128:1.

Openreach GPON is 32:1 with 2 usually kept as spares meaning most is 30:1.
That can be increased to 32:1 when required with permission from those higher up.

They aren't going to change the split ratio when upgrading from GPON to XGS-PON.
Part of the benefit of a PON deployment is that you can upgrade the network just by upgrading either end of the link (the OLT and the ONT).

It would be a fair amount of work increasing the split ratio from 32:1 to 128:1 when upgrading to XGS-PON. I don't see any benefit to doing that either.

They can run both GPON and XGS-PON at the same time.
So that's 2.4Gb/s down and 1.2Gb/s up from GPON and 10Gb/s down/up from XGS-PON.
That's more than enough for many years to come.

By the time bandwidth requirements exceed what XGS-PON can handle the next generation of PON will be released.
There should never be any requirement to increase the split ratio.

Considering the increase in people working from home and CityFibre among others offering increased upload speeds I'd have thought expaning the upload speed profiles available would have been a priority for Openreach as most neither want the expense of a leased line nor need it.

personally 300 down and 100 up would be great for me.

Yes, I'm in an area covered by both OR and CityFibre but I don't want to have another fibre install and house dead kit in my hallway.

I agree. They could I suppose (in theory) introduce a second split at the ODF without too much rearrangement or difficulty. However tis highly unlikely.

I agree with this too.

OR could very easily reduce their asymmetry without too much consequence - if the commercial will (or competitive challenge) was there.

In reply to a post by j0hn83:

In reply to a post by mwarby:

One interesting point is that GPON is split 30 ways, and XGS-PON can be split 128 ways.

Both can be split 128:1
Many networks in Europe do GPON at 128:1.

Openreach GPON is 32:1 with 2 usually kept as spares meaning most is 30:1.
That can be increased to 32:1 when required with permission from those higher up.

They aren't going to change the split ratio when upgrading from GPON to XGS-PON.
Part of the benefit of a PON deployment is that you can upgrade the network just by upgrading either end of the link (the OLT and the ONT).

It would be a fair amount of work increasing the split ratio from 32:1 to 128:1 when upgrading to XGS-PON. I don't see any benefit to doing that either.

They can run both GPON and XGS-PON at the same time.
So that's 2.4Gb/s down and 1.2Gb/s up from GPON and 10Gb/s down/up from XGS-PON.
That's more than enough for many years to come.

By the time bandwidth requirements exceed what XGS-PON can handle the next generation of PON will be released.
There should never be any requirement to increase the split ratio.

the benefit of increasing the split would be that they can reduce the ports needed in OLT to 1/4 of current level. That would be a reduced initial investment and lower ongoing costs (power usage is likely lower)

In reply to a post by Pheasant:

I agree. They could I suppose (in theory) introduce a second split at the ODF without too much rearrangement or difficulty. However tis highly unlikely.

It's what CityFibre are doing. Way cheaper than using twice/quadruple (probably twice in their case) the OLT ports with XGSPON SFP+ modules straight off when they'll be doing virtually nothing. Those optics need line cards, those line cards need chassis, those chassis need space, power and cooling.

It makes a lot of sense to introduce XGSPON and in turn 25GPON / 50GPON on a higher split ratio. Statistical contention will help a ton with capacity planning even when more customers are on the platform and when a segment gets near capacity you aren't splitting a PON with splicing work out in the field as would be needed to split a GPON PON you're decombining them at a headend.

https://cityfibre.com/news/cityfibre-to-upgrade-all-...

The rollout of XGS-PON will allow CityFibre to support more customers on each OLT port, enabling substantial network cost savings, reducing power use across its networks and improving the efficiency of its future network expansion.

Edited by XGS_Is_On (Sun 18-Sep-22 19:25:39)

In reply to a post by j0hn83:

It would be a fair amount of work increasing the split ratio from 32:1 to 128:1 when upgrading to XGS-PON. I don't see any benefit to doing that either.

You don't think using 1/2 or as in your example 1/4 the OLT ports with the resultant savings on optics, line cards and indeed entire OLT chassis and as a result space, power and cooling is a benefit?

In reply to a post by j0hn83:

By the time bandwidth requirements exceed what XGS-PON can handle the next generation of PON will be released.
There should never be any requirement to increase the split ratio.

If requirements exceed what XGSPON can handle wouldn't you be reducing the split ratio, not increasing it?

It's not a lot of work increasing the split ratio for XGSPON. Instead of having an XGSPON optic feeding a single coexistence element directly it goes through a splitter and the outputs go into coexistence elements. To bring XGSPON online engineers would be patching and splicing anyway so it's not a massive deal to introduce a layer of splitting/combining at the headend and the cost savings are considerable. Additional loss introduced by the splitting can be trivially offset through using higher power optics.

Edited by XGS_Is_On (Sun 18-Sep-22 19:15:26)

In reply to a post by XGS_Is_On:

In reply to a post by Pheasant:

I agree. They could I suppose (in theory) introduce a second split at the ODF without too much rearrangement or difficulty. However tis highly unlikely.

It's what CityFibre are doing. Way cheaper than using twice/quadruple (probably twice in their case) the OLT ports with XGSPON SFP+ modules straight off when they'll be doing virtually nothing. Those optics need line cards, those line cards need chassis, those chassis need space, power and cooling.

It makes a lot of sense to introduce XGSPON and in turn 25GPON / 50GPON on a higher split ratio. Statistical contention will help a ton with capacity planning even when more customers are on the platform and when a segment gets near capacity you aren't splitting a PON with splicing work out in the field as would be needed to split a GPON PON you're decombining them at a headend.

https://cityfibre.com/news/cityfibre-to-upgrade-all-...

The rollout of XGS-PON will allow CityFibre to support more customers on each OLT port, enabling substantial network cost savings, reducing power use across its networks and improving the efficiency of its future network expansion.

That's a good point around efficiency, particularly now in light of the energy crisis.

Are other AltNets using XGS-PON splitting at the higher ratios or mostly keeping to 32:1 (or 64:1)?

Its such a shame that Openreach are not agile enough to switch to XGS-PON and may be different split ratios for new deployments rather than continue down the well-trodden path of GPON for something that will clearly need to be done in years to come.

In reply to a post by Pheasant:

That's a good point around efficiency, particularly now in light of the energy crisis.

It's a drop in the ocean though - and you need 6dB more laser power output for each additional 2:1 split.

Remember also that going to 128:1 splits on 10G negates much of the increased bandwidth. Whilst a single 10G PON with 128 users is perhaps a little better than four 2.4G PONs each with 32 users, due to increased statistical spreading; it's still not as good as four 10G PONs each with 32 users.

It seems odd to me that some people are arguing that Openreach should have gone for a Swiss-style model of 1:1 dedicated fibres, and others arguing that they are not aggressive enough with splitting and should have gone with 128:1

ISTM that Openreach are taking a good path here:
* Sell regular GPON services up to 900M (and soon 1G or 1.5G, just for bragging rights)
* Concentrate on getting the fibre into the ground and into people's houses
* The vast majority of users are happy with 40/10 or 80/20 services today anyway
* XGS-PON or 25G or 50G PON are relatively trivial upgrades later down the line, running in parallel on the same fibres
* The bandwidth hogs then don't affect service to the majority

In reply to a post by Pheasant:

That's a good point around efficiency, particularly now in light of the energy crisis.

Are other AltNets using XGS-PON splitting at the higher ratios or mostly keeping to 32:1 (or 64:1)?

Networks tend to be built in the field to 32 or 64:1 but combined at the ODF. They can be trivially split.

Though I've not seen all of them of course, but certainly the two I'm most familiar with are highly unlikely to need to do anything other than decombine and, in time, overlay with newer standards.

In reply to a post by candlerb:

It's a drop in the ocean though - and you need 6dB more laser power output for each additional 2:1 split.

3dB per additional split, insertion loss is minimal. This assuming that it's necessary to hit the receive power range at the ONTs which it almost certainly won't be. Openreach are highly unlikely to have many fibre runs so long that they're at the very edge of the envelope.

The increase in power consumption per module for higher launch power will be negligible and there will be few cases where this is necessary.

Cutting the requirement for additional chassis, line cards, modules and the cooling for that kit is hardly a drop in the ocean: Openreach already have sites that are under pressure due to power and cooling capacity.

In reply to a post by candlerb:

Remember also that going to 128:1 splits on 10G negates much of the increased bandwidth. Whilst a single 10G PON with 128 users is perhaps a little better than four 2.4G PONs each with 32 users, due to increased statistical spreading; it's still not as good as four 10G PONs each with 32 users.

Which is perfectly fine given there'll be hardly any users on XGSPON on these upgraded segments and new XGSPON areas will be a while away from 100% uptake with most customers on cheaper, lower bandwidth tiers. Think you dramatically underestimate the impact of statistical contention, too. A rule of thumb is to provide enough capacity for peak sustained load plus the burst capacity of the highest tier available. On a GPON split that means leaving 40% of it unused. I've seen plenty of cable segments where capacity per premises wasn't changed but moving to fewer, larger upstream channels resolved capacity issues until spectrum could be enlarged or the node split.

It isn't as good as an OLT port for each 32 premises passed for sure, much as an OLT port for each 32 premises passed isn't as good as an OLT port per 8 premises passed.

In reply to a post by candlerb:

It seems odd to me that some people are arguing that Openreach should have gone for a Swiss-style model of 1:1 dedicated fibres, and others arguing that they are not aggressive enough with splitting and should have gone with 128:1

I don't think anyone has said that Openreach aren't aggressive enough with the splitting, only that to do so with an overlay makes sense on many levels. Much as I would appreciate point to point we are where we are and a valid case can be made for following CityFibre in building to 32:1 in the field and combining to something higher near the OLT.

I would actually be surprised if Openreach don't initially combine PONs when they do eventually overlay. They'll be spending a lot of money unnecessarily if they don't. In the UK CityFibre are doing it. AT&T, Verizon and others are doing it internationally, with the two US incumbents using 32:1 for GPON and 64:1 for the XGSPON overlay. Swisscom did it on their retail product when they deployed XGSPON. 64:1 seems to be the sweet spot right now and standard practice.

Thanks. I think we can all agree the general 1:32 split in the field is very unlikely to change. Any further increases in splitting density (if desired) would be at the ODF where as you say it is relatively simple / trivial to achieve.

In reply to a post by Pheasant:

Thanks. I think we can all agree the general 1:32 split in the field is very unlikely to change. Any further increases in splitting density (if desired) would be at the ODF where as you say it is relatively simple / trivial to achieve.

Agree and will add that increasing the split is desirable. Openreach work in strange and mysterious ways of course but the experience of AT&T, Verizon and other ILECs might indicate the way forward.

Increasing the split ratio is trivial. More to the point returning it to the original field ratio is also trivial, apart of course from the OLT ports and all the expense potentially associated with provisioning those ports. It's not a lift and shift, it's on top of the existing GPON infrastructure. Some reuse of chassis and line cards where they are capable of both but it's largely going to be new cards and in turn chassis as the card requirement overflows the capacity of the existing chassis.

Where they've dual-capable line cards there's even more of an incentive to use as few ports as possible. Saving as many ports as possible on XGSPON reduces the number of line cards that'll be needed to deliver XGSPON and in turn is more likely to make the difference between needing new chassis or not. When you can use those unoccupied ports on the existing line cards every port counts before you start considering every card then every chassis.

In reply to a post by XGS_Is_On:

In reply to a post by Pheasant:

Thanks. I think we can all agree the general 1:32 split in the field is very unlikely to change. Any further increases in splitting density (if desired) would be at the ODF where as you say it is relatively simple / trivial to achieve.

Openreach work in strange and mysterious ways of course but the experience of AT&T, Verizon and other ILECs might indicate the way forward.

Not really. Openreach is a gigantic company (especially when you think of them as part of BT Group, and they are reliant on other parts of the group), so it’s unfair to compare them to the newer Altnets.

In reply to a post by _Icaras_:

Openreach is a gigantic company (especially when you think of them as part of BT Group, and they are reliant on other parts of the group)

I think BT group were a gigantic company when they had 250,000+ employees and had the whole market to themselves but they aren't that same company anymore.

They are the same in terms of how they are regulated though, in terms of the rules imposed on them for wholesale access, fault fixes etc. Openreach wouldn't be able to run a network as congested as some of Virgin Media's HFC areas for as long as VM let faults run on for without Ofcom getting involved.

In reply to a post by jpm:

They are the same in terms of how they are regulated though, in terms of the rules imposed on them for wholesale access, fault fixes etc. Openreach wouldn't be able to run a network as congested as some of Virgin Media's HFC areas for as long as VM let faults run on for without Ofcom getting involved.

I'm sure you're right but I was just referring to the gigantic comment not how they are regulated

In reply to a post by _Icaras_:

In reply to a post by XGS_Is_On:

Openreach work in strange and mysterious ways of course but the experience of AT&T, Verizon and other ILECs might indicate the way forward.

Not really. Openreach is a gigantic company (especially when you think of them as part of BT Group, and they are reliant on other parts of the group), so it’s unfair to compare them to the newer Altnets.

Openreach isn't gigantic compared to AT&T and Verizon and they aren't altnets. ILEC = Incumbent Local Exchange Carrier. AT&T and Verizon = 2 of the companies the USA equivalent of BT, Bell, was split into when it was separated geographically. AT&T plan on reaching 30 million premises with full fibre by 2025, about the equivalent of the entire UK, and Verizon presently have full fibre to 17.9 million homes having sold off part of their footprint to Frontier Communications, who also actually have an XGSPON overbuild program underway that also increases the split ratio.

Verizon's market capitalisation is currently a little over $171 billion, AT&T's a little over $118 billion. It seems fair to suggest they aren't altnets compared to whom Openreach are 'gigantic' given BT Group's market cap is about $15 billion.

Edited by XGS_Is_On (Tue 20-Sep-22 17:03:49)

The Altnet comment did make me chuckle just a little. 😂

In reply to a post by XGS_Is_On:

Verizon presently have full fibre to 17.9 million homes

One of my US friends is on Verizon FiOS FTTP in New Jersey. Verizon installed poles all down his long road and strung the fibre. He was happy, but its only 600 Mbps down 50 Mbps up, much more reliable than his old Coax cable TV service from Comcast.

In reply to a post by jchamier:

In reply to a post by XGS_Is_On:

Verizon presently have full fibre to 17.9 million homes

One of my US friends is on Verizon FiOS FTTP in New Jersey. Verizon installed poles all down his long road and strung the fibre. He was happy, but its only 600 Mbps down 50 Mbps up, much more reliable than his old Coax cable TV service from Comcast.

Strange. They've no asymmetric packages on their rate card. Your friend may need to have a conversation with them.

Yeah he’s probably mistyped on his email to me. 😂