g.Fast in the countryside

First off I know what a g.Fast pod looks like having seen loads on cabinets on the Bridgeton exchange in Glasgow. So a couple of weeks ago I was driving in Dundee and notice a few g.Fast cabinets. A quick check on Streetview suggest for example that cabinet 26 on Brook Street is one. Rather odd because Dundee is not on the list of places to be getting g.Fast.

So today on my way home I stopped off to get some milk at the Coop in Tayport. It just happens to be right next to cabinet number two for the Tayport exchange (ESTAY), and blow me down it's not sporting a g.Fast pod. Now Tayport is as things go firstly in the countryside, secondly small, according to SamKnows supports about 1900 residential properties and approx 50 none residential properties. It also only has to the best of my knowledge four cabinets. Three original ones, and a more recent BDUK one at the exchange to deal with exchange only lines. This has the number four stenciled on and being recent is presumably the highest numbered one.

This of course prompted an immediate trip around the other cabinets which reveals that one, two and three are all supporting g.Fast pods, but the brand new one at the exchange (number four) is as of yet not.

A quick check on the BT Wholesale checker, and there is no g.Fast availability on my line, but I know on the Bridgeton exchange in Glasgow it took some pods best part of a year to show availability. I guess it's possible that I am too far away, but then again I currently sync at something around the full 80/20. Though for reasons I don't understand the HG612 is reporting much higher SNR, attenuation and power down than up. Specifically I get speed 77.50/22.98, SNR 12.50/2.6, attenuation 17.60/0.00, power 13.0/7.3. Anyway I am not entirely sure the route taken there is a pole in the back garden and the cable seems to go over the wall and into the ground in the church grounds. My guess it goes to one of the chambers in the footpath between the church and the cabinet (there are several), and is probably around 200m from the pole to the cabinet. So potentially in range of a g.Fast service. That said my guess is that only a small percentage of Tayport will be able to get a gFast service. Which makes the deployment of the pods even more strange.

Anyway does anyone know anything about these unadvertised gFast pod deployments? Also does anyone know when Openreach started deploying g.Fast outside cities?

Finally for anyone who thinks I am mistaken here is cabinet two on Streetview which quite old as it does not show the VDSL fibre twin.

https://goo.gl/maps/QV1cAHEA9hv

Here is the photo I took today of the same cabinet. It has now been reskinned, and supports a fibre twin in addition to the gFast pod.

https://small.buzzard.me.uk/stuff/tayport2_gfast.jpeg

https://labs.thinkbroadband.com/local/broadband-map#... shows a few more rural location pods too

So you don't have to be in a city though so far the majority is.

It's very odd if an HG612 is reporting an upstream sync of 22.98Mbps.

Are you looking at the GUI or the telnet figures?

Edited by RobertoS (Sun 07-Oct-18 17:36:47)

Here is the truncated output of xdslcmd info --show. What I don't understand is max downstream is 77264 Kbps, but the downstream rate on bearer 0 is 79999 Kbps. How does that work. It's Huawei cabinet.
# xdslcmd info --show
xdslcmd: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 0
Last initialization procedure status: 0
Max: Upstream rate = 22974 Kbps, Downstream rate = 77264 Kbps
Bearer: 0, Upstream rate = 19999 Kbps, Downstream rate = 79999 Kbps
Bearer: 1, Upstream rate = 0 Kbps, Downstream rate = 0 Kbps
Link Power State: L0
Mode: VDSL2 Annex B
VDSL2 Profile: Profile 17a
TPS-TC: PTM Mode(0x0)
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 2.6 12.6
Attn(dB): 17.6 0.0
Pwr(dBm): 13.0 7.3

In reply to a post by jabuzzard:

Anyway does anyone know anything about these unadvertised gFast pod deployments? Also does anyone know when Openreach started deploying g.Fast outside cities?

Openreach do intend to cover up to around 6 million premises with G.fast by 2020. This has recently been scaled back from 10 million as they have decided to focus more on rolling out FTTP (going from a target of 2 million to 3 million).

There are no announcements to where they are currently deploying this technology besides pilot areas. But as someone who is not in a pilot area, I can confirm that they are in fact rolling out G.fast outside of cities as I have seen quite a few cabinets with pods fitted and, in some cases, reshelled prior if necessary. I am in an area that had FTTC as part of BDUK, so I was surprised to see any commcercial interest in the area. Unfortunately, it can take a long time to actually enable the pods since cable links need to be upgraded.

I am not fully aware of everything what would make an area a good candidate for G.fast, but I imagine that having dense premises is preferable to maximise the premises covered with G.fast. Since G.fast uses 19-106 MHz as opposed to the approx 0.14-17 MHz that VDSL2 17a uses, the distance is a limiting factor; it is said to be around approx 370 meters before the speed drops off below 100 Mbps (at which point ISPs won't sell it to you as far as I know). But I would say at the top-end of the distance the upload is likely going to be noticeably less than with FTTC/VDSL2 17a which is currently employed.

In reply to a post by jabuzzard:

Here is the truncated output of xdslcmd info --show. What I don't understand is max downstream is 77264 Kbps, but the downstream rate on bearer 0 is 79999 Kbps. How does that work. It's Huawei cabinet.
. . .
Max: Upstream rate = 22974 Kbps, Downstream rate = 77264 Kbps
Bearer: 0, Upstream rate = 19999 Kbps, Downstream rate = 79999 Kbps
Bearer: 1, Upstream rate = 0 Kbps, Downstream rate = 0 Kbps

The max rate is simply what the modem believes the line would be capable of at the current target SNRM. The max rate for the downstream is less than the current because the conditions would not favour a high downstream speed if a resync took place.

Anyway, your upstream rate is 19999 Kbps, therefore essentially 20 Mbps and not almost 23 Mbps. Though the indication is, as mentioned above, that modem thinks it could achieve almost 23 Mbps at the current target margin if there was no cap.

Edited to fix typo.

Edited by deleted (Sun 07-Oct-18 21:10:37)

Ah, that makes more sense .

Bearer 0 is your actual sync speed. Max (attainable) is what your line could theoretically reach if it wasn't capped at the Openreach product limits of 80/20.

The presence of Bearer 1 shows that G.INP is active, with more details in the parts you have cut off. It shows as 0 speeds as what the stats show is rounded to kbps, but Bearer 1 connects at 200bps.

So you are connected at the full 80/20.

Note that the SNRM figures are Down/Up, the opposite of the speed ones. We would need to see the same stats taken immediately after the next re-sync to see what their targets are, but the fact that your Max Attainable down is so much lower that Bearer 0 shows that considerably more noise is present than was the case when you connected. The upstream on the other hand is remarkably high but not a problem at this stage.

(Note, never use the HG612 GUI for stats in you think of doing so. The three lines in the GUI for SNRM, Attenuation and Power have one line of figures omitted and one duplicated, whilst the labels appear correct).

Edited by RobertoS (Sun 07-Oct-18 20:14:46)

In reply to a post by RobertoS:

Bearer 0 is your actual sync speed. Max (attainable) is what your line could theoretically reach if it wasn't capped at the Openreach product limits of 80/20.

In this case, yes, as the sync speed is already max of what the product allows. Although the max attinable can be applied to any connection regardless of whether it is at the cap or not. It's just a theoretical max that the modem thinks is possible with current conditions and parameters.

With interleaving in place, the case is slightly difference. As j0hn83 would say:
The max attainable is exaggerated with interleaving.

As someone who bounces around Dundee and Perth a lot. I can say that Perth had gfast pods fitted around 6 months ago. The gfast cabinets in Dundee have only appeared in the last month or so and seems to have included almost every PCP in the city centre although I have seem some in outlying areas too.

As you say there is no mention of gfast on the checker although I did spot a gfast record showing "planned" for another area of the city but I can't remember what the postcode was.

There is also a flurry of FTTP happening in rural areas too which seems to be also "off the record"

I was replying to the specific case, and the specific question and the poster�s misunderstanding of the basic meaning of �Max� that led to the telnet stats being posted.

Though for reasons I don't understand the HG612 is reporting much higher SNR, attenuation and power down than up. Specifically I get speed 77.50/22.98, SNR 12.50/2.6, attenuation 17.60/0.00, power 13.0/7.3.

Seeing as you seem to think I gave insufficient information about the general case, you forgot to point out that on FTTC, interleaving without G.INP active also lowers the sync speed and adds at least 8ms to latency. That the removal of interleaving raises the sync at the same time as it lowers the attainable and removes that additional latency.

Plus of course G.INP does have a small amount of interleaving, but without adding to latency and without �exaggerating� the attainable. It also has an even lower amount of interleaving on Bearer 1.

Various other difference can be seen with the full stats of course, particularly wrt G.INP. As I haven�t covered them, perhaps you would like to.

Edited by RobertoS (Sun 07-Oct-18 23:06:33)

In reply to a post by RobertoS:

I was replying to the specific case, and the specific question and the poster�s misunderstanding of the basic meaning of �Max� that led to the telnet stats being posted.

Yes, I can see the intention now. I was initially unsure of whether it was intended as a specific or general response. Because of forward thinking, I decided to employ clarity in a general case in case there is a misunderstanding if this thread is stumbled upon in the future. Perhaps unnecessary, however.

In reply to a post by RobertoS:

Seeing as you seem to think I gave insufficient information about the general case . . .

That is not the case at all, but rather an assumption of yours. In reference to the interleaving, I even thought this information would be a bit unncessary at the time, but I just wanted to quote a knowledgeable forum member but not go into much detail at the time myself. I understand you tried to make light out of the situation in the form of a cheeky jab, which is fine. But you should know that I am not doubting your knowledge but rather just trying to aid clarity.

Keep digging ..

I can�t stop you.

In reply to a post by RobertoS:

Keep digging ..

I can�t stop you.

Okay.

@mods, this string of comments should be pruned as clearly serve no purpose. Rather, they are intended to provoke a response which would be off-topic, counter-productive and a waste of time.

I noticed the Perth city centre pods for the first time last week. I tried a number connected to cab 14 through the checker but no gfast products were offered yet. I'd love to see FTTP reaching out into Perthshire.

In reply to a post by j0el:

Openreach do intend to cover up to around 6 million premises with G.fast by 2020. This has recently been scaled back from 10 million as they have decided to focus more on rolling out FTTP (going from a target of 2 million to 3 million).

So overall, a nett reduction of 3 million premises with access to Ultrafast broadband

No-one seems to particularly care, the extra 1 million full fibre premises was what got people excited.

I wonder what the cost of a single G.Fast cabinet is compared to rolling out FTTP availability/infrastructure to the same area that the G.Fast has a useful effect on. If the final installation from the DP to the inside of the premises were done on a fixed cost basis like a simple phone line is, plus an affordable price for the on-premises ONT. Does anyone know the cost of the ONT please?

Then we extend that calculation to wider areas.

In effect I'm questioning whether the current economic model used by Openreach and BT Group is the right one. Is time and money being wasted on FTTC infill cabinets and other non-FTTC areas being provided with FTTC? Including the planning aspects where those resources could be made available for FTTP instead.

In many places in Cornwall FTTP delivered over telegraph poles is rife. Such poles being prevalent in other rural areas. Aggregation nodes could be fed that way as well of course.

In reply to a post by RobertoS:

If the final installation from the DP to the inside of the premises were done on a fixed cost basis like a simple phone line is

It already is - the wholesale price is £92. There's no extra charge for the ONT.

The cost of ONT (roughly £50) is quickly recovered through rental. A port on the OLT is more expensive - maybe £1,000 - but is shared by all users on the same splitter and amortised over several years.

The *real* cost of making FTTP available in the backbone is in pulling fibres, unblocking ducts (or laying new ones where they are full), terminations etc.

The trouble with rolling out FTTP to G.fast capable areas (i.e. those within 300m of the cabinet) is that these people already get the max 80/20. They may pick up some customers who want faster speeds than that, but it increases the digital divide for everyone else.

Is time and money being wasted on FTTC infill cabinets and other non-FTTC areas being provided with FTTC?

With access to a mountain of paperwork impossible to say.

What we can say is this...

Not all infill with BDUK projects is with VDSL2 anymore, and seeing some sub 24 areas with existing VDSL2 get FTTP from the projects.

The pattern from Fibre First (commercial) is at present in Northern Ireland, lots of VDSl2 overlay, but in England so far the bulk of what has been found is EO to FTTP. VDSL2 cabinet overlay in the cities does appear on the way though.

When doing FTTP and cable, the ones most often missing out are flats in the roll-outs that are on-going

Thanks candlerb and Andrew .

Edited by RobertoS (Mon 08-Oct-18 16:15:05)

In reply to a post by candlerb:

The cost of ONT (roughly £50) is quickly recovered through rental. A port on the OLT is more expensive - maybe £1,000 - but is shared by all users on the same splitter and amortised over several years.

The *real* cost of making FTTP available in the backbone is in pulling fibres, unblocking ducts (or laying new ones where they are full), terminations etc.

The trouble with rolling out FTTP to G.fast capable areas (i.e. those within 300m of the cabinet) is that these people already get the max 80/20. They may pick up some customers who want faster speeds than that, but it increases the digital divide for everyone else.

The cost of a OLT per subscriber is quite low, exactly how low depends on the split rate, but at 16 subscribers per OLT port you are looking at under £20 per subscriber and that is retail prices and I would expect Openreach as a volume customer to be paying significantly less.

Note that SFP based ONT's are $20 on Aliexpress/eBay. For example here is one on eBay shipped to you for just $25 and it's a Huawei so basically the SFP version of the ONT that Openreach are currently using Huawei HPSP2120. Frankly a much better idea than what they are using in my view as there are a whole range of routers with SFP ports allowing the end user to save a box and use whatever router they want. If needs be a media converter is only ~£25 retail so less than many dedicated ONT's.

I guess the issue with g.Fast pods in less dense areas than say some of the trial areas (for example Glasgow Bridgeton) is that percentage of properties on the cabinet able to get a useful speedup is going to be significantly lower. I would estimate that better than 90% of properties of cabinets in Bridgeton will get a gFast service with a much lower figure for Tayport, especially for cabinet two, which means doing a mix of gFast and FTTP, which can't be great.

I believe the ONT being referred to in Bobs post is Optical Network Termination.

The item you have linked to is an optical splitter.

It's not a splitter.

It's a rather strange item: a GPON interface in SPF form factor. You can't plug it into a switch or router, as it says:

Please note this GPON ONU module is GPON to GPON protocal ! ( not GPON to Ethernet )

I guess it's used in some sort of ONT with interchangeable optics.

Anyway, an OpenReach-supplied ONT is not a router. It's like a transparent modem - you talk PPPoE to it - and you still need a router of your own.

In reply to a post by candlerb:

Anyway, an OpenReach-supplied ONT is not a router. It's like a transparent modem - you talk PPPoE to it - and you still need a router of your own.

And long may it continue that way! Some FTTP Operators - such as Gigaclear Residential - force you to use a combined ONT/router which can potentially create issues such as double NAT if trying to hook up your own router to their equipment. At least with Openreach, their ONT is just a dumb light converter ( ie does zero routing) and you're free to use any router you like.

Apologies then, you�ll admit it does look exactly the same as a splitter though.

I knew your last point though.

Since when has a splitter ever looked like an SFP? But yeah on second looks the description is somewhat odd. However if you want to take this post on face value then it does indeed work in at least one specific router.

https://community.ubnt.com/t5/EdgeRouter/Edgerouter-...

In reply to a post by jabuzzard:

However if you want to take this post on face value then it does indeed work in at least one specific router.

If it is indeed a whole ONT-in-an-SFP that works in an ethernet SFP slot, that raises a bunch of other questions: how do you enrol it to the OLT? Which vendors of OLT does it work with? Is it manageable from the network side?

Post deleted by Zarjaz

I think you guys must be talking about different links

Well the standard defines three ways in which an ONT can register/authenticate with an OLT. The first is to use just the serial number of the device, the second is to use a serial number and password and the third is to use a password. Openreach are by all accounts just using the serial number of the ONT. As such if you have the ability to change the serial number of your ONT to match the one supplied by Openreach it will work.

There are a number of people who have achieved this and one day a ONT vendor is likely to come up with an ONT that explicitly supports it. Here is for example how to change the serial number on a Ubiquiti ONT to whaterver you want

https://blog.onedefence.com/changing-the-gpon-serial...

Kind of cool as the Ubiquiti ONT can be PoE powered.

In reply to a post by baby_frogmella:

And long may it continue that way! Some FTTP Operators - such as Gigaclear Residential - force you to use a combined ONT/router which can potentially create issues such as double NAT if trying to hook up your own router to their equipment. At least with Openreach, their ONT is just a dumb light converter ( ie does zero routing) and you're free to use any router you like.

If you are willing to hold your nose while doing it, you can mostly overcome this sort of stupidity using proxy ARP. It will however require you use a real router, say Cisco, Juniper, pfSense, Mikrotek, Ubiquiti, VyOS, DD-WRT etc. The use of consumer junk from the likes of Netgear, Asus etc is extremely unlikely to work. It's not quite bridging mode as the hunk of junk you are trying to bypass still does all the NAT and any port forwarding, but otherwise it is out of the picture. Better than double NAT in my view. I won't try and explain it because it would be over the heads of 99% of people here as it's a bit of esoteric enterprise grade networking.

Personally I would be cracking open the Gigaclear router and looking for any serial port headers so at the very least I could insert a custom route if not put it into bridging mode. A quick poke on the internet shows that under the hood it like most stuff is running Linux so bridging mode is technically possible.

Though I might well start by trying a out with a Ubiquti ONT and getting that to pretend to be a Gigaclear ONT seems as though it is by far the ONT best documented hack of this type and one which does not require breaking out the soldering iron.