FTTP vs FTTC and peak time throttling

Hi. First post for a while.

I used to be quite clued in on all this stuff, pre-fibre, but now I'm very rusty, so I'm after some general wisdom, please.

Once upon a time, there were 2 types of ISP: one would charge more and give you unfettered access to bandwidth; and one would be cheaper and implement peak period throttling. This was because Linx data pipes were expensive, and it was cheaper to share them among more customers. I've always gone for "unlimited" DSL, O2, then Sky, then Zen. I'm currently getting 48Mbps down, with no noticeable slowdown at peak periods.

Fast forward, and a guy from Trooli comes knocking at my door. He's selling FTTP, new to our semi-rural area (Dorset). Now, I'm interested, because I know we'll have to go FTTP eventually, and, well, who doesn't want faster internet?

But my question is (eventually): never mind the headline connection speed at the premises (150M/500M/900M/2G), what sort of minimum speeds would I expect when all of their customers are streaming and downloading in the evenings, or is that not a thing anymore?

Edited by YammerUK (Thu 29-Aug-24 08:23:52)

FTTP is a shared system, I think it is shared between 32 houses, that is Openreach, so altnets share between less. Will it make a difference to you? No, that was one of my worried before I went to FTTP.

I can't remember the amount of bandwidth that is shared, but the chance of it getting saturated is pretty slim unless every one of those 32 properties use the full bandwidth they can at the same time, and that is not going to happen. I would not worry about it, been there, done that.

That's great information, thanks.

I see what you're saying about premises sharing connection equipment, similar to FTTC premises sharing an exchange DSL board. But what about the ISP end?

In the old days, you were limited by the amount of bandwidth the ISP could allocate per customer at peak periods, not the connection speed to the local exchange. So, it didn't matter what your headline "speed" was, you'd get throttled further down the line between the ISP and the LINX.

It's not a thing, you don't need to be concerned by it. Some ISPs will have minimum speed guarantees listed along with the headline package but I can't imagine a small provider like Trooli has enough customers for capacity problems to creep in.

Bandwidth is essentially unlimited these days; London is one of the cheapest places in the world to buy Internet transit.

Very large providers don't peer at the LINX (or equivalent exchange points like LONAP) - they have dedicated private peering links between themselves, which are as large as needed.

For smaller providers peering at exchange points, they just pay for whatever port size they need. Most providers are not going to be choking here, although a badly-run network which has not done its capacity planning properly might do.

Is Trooli a well-run network or a badly-run network? Who knows. They might simply buy blended transit from an upstream provider, and the question then becomes whether *that* is a well-run or badly-run network.

One thing that's important to understand is that on a retail-only altnet you have no choice of provider: you get whatever service they provide. For comparison: if you had a connection via a wholesale network like Openreach or Cityfibre, you get a range of ISPs who compete on price and service, and you can easily switch between them. (Well, perhaps not *easily* in the case of Cityfibre, but that's a different story).

A few days ago Trooli announced a wholesale partnership with Zen - so there may be *some* choice of ISP on their network forthcoming, albeit probably not immediately.

Otherwise, if you're considering the Trooli retail offering, check whether the service meets your needs. Do you know or care what CGNAT is, or static IP, or IPv6? If you don't, then their service may be fine for general browsing and downloading. If you do, then check more carefully what service they provide. In any case, check what the contract terms are. Check whether there is compensation in the case of an extended outage.

If reliability is important, you could always keep the FTTC connection running in parallel with Trooli for a period of time, until you're sure it's good enough to terminate the FTTC.

In reply to a post by YammerUK:

That's great information, thanks.

I see what you're saying about premises sharing connection equipment, similar to FTTC premises sharing an exchange DSL board. But what about the ISP end?

In the old days, you were limited by the amount of bandwidth the ISP could allocate per customer at peak periods, not the connection speed to the local exchange. So, it didn't matter what your headline "speed" was, you'd get throttled further down the line between the ISP and the LINX.

His information was wrong about Openreach sharing more than altnets, their ratio is as low as it gets, but broadly okay. There was actually a lot more sharing of bandwidth on his beloved FTTC where 300+ homes could share a gigabit than there is on full fibre where the most homes sharing is 128 and they share about 8.5 Gbit. The lower standard it's about 2.5 Gbit between 30-32 homes.

ISPs aren't the bottleneck anymore. Advertising and regulatory changes force them to advertise the average speed during peak times. If you aren't getting full speed at peak times there's a problem somewhere.

Just FYI the old school issues were when everyone besides the cable company was paying BT Wholesale for bandwidth from the exchanges to the ISP. That cost a massive amount and was why some throttled. They were paying seven figures a year for less than a gigabit of capacity. Ouch!

In reply to a post by zyborg47:

FTTP is a shared system, I think it is shared between 32 houses, that is Openreach, so altnets share between less. Will it make a difference to you? No, that was one of my worried before I went to FTTP.

I can't remember the amount of bandwidth that is shared, but the chance of it getting saturated is pretty slim unless every one of those 32 properties use the full bandwidth they can at the same time, and that is not going to happen. I would not worry about it, been there, done that.

Perhaps you can provide evidence of this .
As the OP is asking about a particular Alt Net , then referencing OR serves what purpose ?, apart from your almost pathological need to denigrate them

Don't forget the OP is currently on FTTC. Which is Openreach from the premises to the exchange. Backhaul may be BT Wholesale or Zen itself.

I think the comparison is valid.

Where's the denigration of OR in the post? Or do you have a "pathological need to denigrate" him?

In reply to a post by pluralist:

Don't forget the OP is currently on FTTC. Which is Openreach from the premises to the exchange. Backhaul may be BT Wholesale or Zen itself.

I think the comparison is valid.

Good for you , you can think what you like .

Where's the denigration of OR in the post? Or do you have a "pathological need to denigrate" him?

I underlined the part in question ( perhaps you missed it ) but for clarity ‘so alt nets share between less ‘ , a poorly constructed sentence but the meaning is clear , that Alt Nets use a ‘better’ splitter ratio that OR’s 32/1 , and as splitters are built from -3dB 1-2 ‘splits’ , 1 to 2 , 2 to 4 , 4 to 8 , 8 to 16 and 16 to 32 etc, it follows it would have to be a split of , 2 ,4 , 8 or 16 , the poster gives no citation to support this assertion that Alt Nets use a better ratio, it’s simply more of their jaundiced view that in comparison to Alt Nets , OR must be worse , perhaps some do use a different split ratio , for all the ( especially on here ) wet dreams about XG-PON and XGSPON, it can ( by design) use a much larger split ratio of 1-128 , some Alt Nets may even be point to point , but that’s not what that particular poster claimed, if they had posted ‘Alt Nets may share between less ‘ still a terrible sentence, but one I wouldn’t have commented on.

As far as myself having a pathological need to denigrate him , it’s the first and only time I’ve referred to that poster in any of my contributions , for you to deny that they comment continually and negatively about OR , ( and BT ) even in cases like this where the subject wasn’t even OR in the first place, would be laughable

Edited by Iniltous (Thu 29-Aug-24 14:36:48)

In reply to a post by pluralist:

Don't forget the OP is currently on FTTC. Which is Openreach from the premises to the exchange. Backhaul may be BT Wholesale or Zen itself.

True, but that has what to do with PON split ratios?

In reply to a post by pluralist:

I think the comparison is valid.

Where's the denigration of OR in the post? Or do you have a "pathological need to denigrate" him?

The comparison was wrong, no altnet has a lower split ratio than Openreach's standard one or a more conservative capacity planning criteria. Given the OP doesn't have Openreach available to them it was also irrelevant. The split is less important than capacity planning criteria and ease of upgrade.

I rarely agree with Initious but the pathological need to denigrate Openreach is an ongoing theme in the man's posting, right down to the inane nicknames he uses for Openreach et al. I can't say I'm surprised if there's sensitivity and that's without being able to read whatever nonsense is written in The Park.

In reply to a post by XGS_Is_On:

The comparison was wrong, no altnet has a lower split ratio than Openreach's standard one or a more conservative capacity planning criteria.

Possibly the point was that Altnets have very few customers. If they planned for 100% coverage at a 32:1 split ratio, you're unlikely to have more than 3 or 4 customers on one PON - a bit more for Cityfibre. But then again, Altnets have probably already taken into account that they're unlikely to exceed 20-25% takeup and have designed the network accordingly. CityFibre have apparently hit capacity limits in a few areas.

Conversely, Openreach FTTP will eventually have close to 100% take-up when they migrate all existing customers off copper. But then again, Openreach's customer base includes lots of "normal" people who use broadband for nothing more than the odd phone call, browsing Amazon and the odd SD TV stream, so their impact on contention is effectively zero.

Openreach do ‘dimension’ their network to include every address in the PON area , but even the most optimistic OR employee isn’t expecting anything like 100% take up , possibly on a ‘newsite’ where no alternative is available, then there may be a higher take up rate , but many newsites have VM available from the outset as well as OR .

Although probably a little over simplistic, OR state a take up rate of around 30% , they have a practical limit of 30 users ( not 32 ) per splitter , so the chances are currently someone in a retro build area ( so copper also exists ) a PON/splitter is likely to have no more than around 10 users , obviously this may increase as copper pair users are encouraged to change to FTTP , but conversely customers may have multiple networks available ( Virgin , Alt Net ) so may actually decrease, so 100% take up is not very likely at all , TBH , I doubt poster that stated Alt Nets have a better split ratio was being that nuanced, it’s just their preconceptions of OR

Edited by Iniltous (Fri 30-Aug-24 10:33:20)

In reply to a post by candlerb:

Conversely, Openreach FTTP will eventually have close to 100% take-up when they migrate all existing customers off copper.

😂Around this neighbourhood the question will be if OR have any customers left when they roll out FTTP. The "cherry" has already been picked by alt net and VM.

In reply to a post by candlerb:

Possibly the point was that Altnets have very few customers. If they planned for 100% coverage at a 32:1 split ratio, you're unlikely to have more than 3 or 4 customers on one PON - a bit more for Cityfibre. But then again, Altnets have probably already taken into account that they're unlikely to exceed 20-25% takeup and have designed the network accordingly.

I'll let my main ISP know: they're at 45% take up on my PON with a 1:128 ratio.

In reply to a post by jchamier:

Around this neighbourhood the question will be if OR have any customers left when they roll out FTTP. The "cherry" has already been picked by alt net and VM.

How do you know this? Where I live there are 2 altnets, no VM, and no sign of OR FTTP. I would love to know what the local take-up of FTTP from the altnets is, but short of walking round and trying to spot splice boxes on walls I don't know any way to find out.

Ask Trooli what speed you can expect all the time. If they don't provide this speed follow the Ofcom code of practice.

FTTP ISPs sometimes use the connection speed which will be faster than the max rate that data can be sent or received because of the over heads.

If all you are doing is streaming some TV, browsing the internet and using email I doubt that you will notice a radical improvement.

The Trooli price for £150Mbps is less that I am paying for 40/10 FTTC! At a differernt I have had a Gigaclear FTTP service which has been fine for the past 9 years.

In reply to a post by Thaumaturge:

How do you know this? Where I live there are 2 altnets, no VM, and no sign of OR FTTP. I would love to know what the local take-up of FTTP from the altnets is, but short of walking round and trying to spot splice boxes on walls I don't know any way to find out.

I walk the neighbourhood and spot the VM boxes on the walls that accelerated after the first lockdown in 2020 when home working became a 'big thing'. My own FTTC speeds had been slowing fast, I moved to VM in 2019, because my original 60/4 service became a 40/2 service and I was fed up with slow upload. (Poor copper in my block of flats hindered the upload. Downstream reduction was described to me as likely cross talk as more and more people moved from 40/10 to 80/20 services as prices fell).

Our relatively new alt net is overhead, and has installed telegraph poles in roads that didn't have them before, so that is very easy to spot whom is connected as both OR and VM are underground.

I walk the neighbourhood and spot the VM boxes on the walls...

Mmm, much as I would have to do. The TBB broadband stats are, ahem, not very accurate around where I live. The broadband map only goes down to postcode granularity, and while most postcodes show as having at least one connection, they cover typically 30 addresses, so that might represent anything from 3% to 100% takeup.

One altnet did PIA here. The other dug up about 90% of the village and installed Toby pots. My connection is with them, but I live in the older part of the village where there were existing poles so I'm in the minority with an overhead connection from an OR pole. The OR copper is still there too, which makes the fibre connections less easy to spot.

In reply to a post by Iniltous:

Perhaps you can provide evidence of this .
As the OP is asking about a particular Alt Net , then referencing OR serves what purpose ?, apart from your almost pathological need to denigrate them

Ok, lets put it a different way, from reading around before I got zzoomm, I read that Openreach a contention ratio is 32-1, while Zzoomm has 20-1, so it may not be the same for all networks I suppose.

The info is not the easiest to get, I got from different forums and sites, and it seems that they match up with each other.
Oh do you feel superior using words like denigrate?

No, this is nothing to do with my hatred for OR and BT.

In reply to a post by YammerUK:

That's great information, thanks.

I see what you're saying about premises sharing connection equipment, similar to FTTC premises sharing an exchange DSL board. But what about the ISP end?

In the old days, you were limited by the amount of bandwidth the ISP could allocate per customer at peak periods, not the connection speed to the local exchange. So, it didn't matter what your headline "speed" was, you'd get throttled further down the line between the ISP and the LINX.

at the end of the day it will not make any difference to you, you will get the speed you pay for.

In reply to a post by zyborg47:

In reply to a post by Iniltous:

Perhaps you can provide evidence of this .
As the OP is asking about a particular Alt Net , then referencing OR serves what purpose ?, apart from your almost pathological need to denigrate them

Ok, lets put it a different way, from reading around before I got zzoomm, I read that Openreach a contention ratio is 32-1, while Zzoomm has 20-1, so it may not be the same for all networks I suppose.

The info is not the easiest to get, I got from different forums and sites, and it seems that they match up with each other.
Oh do you feel superior using words like denigrate?

No, this is nothing to do with my hatred for OR and BT.

If you read somewhere that your ISP ( and as your initial reply stated not just up your ISP but all other Alt Nets use a ‘better’ split ratio ) provide some evidence to support that claim , FWIW , 20-1 split doesn’t lend itself to the way optical splitting works , so presumably your claim is either that your ISP takes a 32-1 splitter and disables output fibres 21-32 , or they have some magical way to get to 20-1, in which case it would be interesting to know of this technology, BTW , there is nothing wrong with saying that’s unverified nonsense unless you can point to where you obtained this information.

Why does a word like denigrate bother you ? , I could have stated , ‘why do you always slag off Openreach and BT ‘ , it conveys the same message, but I think it’s a juvenile , aggressive and disrespectful way to address anyone, even you , so I don’t do it , would you prefer to be addressed in a disrespectful way ?

I’m not sure you realise the irony, you are effectively saying ‘my hate for BT and Openreach has nothing to do with my negativity towards them’ , in this case ( even though they were not the subject of the discussion ) you included a negative reference for what purpose ? , the OP apparently hasn’t the option to use OR anyway

Edited by Iniltous (Sat 31-Aug-24 10:35:14)

I'd be amazed if someone was building an XGS-PON network on a 1:20 split ratio

In reply to a post by XGS_Is_On:

ISPs aren't the bottleneck anymore. Advertising and regulatory changes force them to advertise the average speed during peak times. If you aren't getting full speed at peak times there's a problem somewhere.

Just FYI the old school issues were when everyone besides the cable company was paying BT Wholesale for bandwidth from the exchanges to the ISP. That cost a massive amount and was why some throttled. They were paying seven figures a year for less than a gigabit of capacity. Ouch!

OK. That clears things up. Thanks.

In other news: I contacted my current ISP, Zen (who have generally been pretty good), told me that they are negotiating use of Trooli's FTTP infrastructure, so I may get to stay where I am anyway!

In reply to a post by YammerUK:

Zen (who have generally been pretty good), told me that they are negotiating use of Trooli's FTTP infrastructure

Some info:
https://www.zen.co.uk/blog/posts/zen-blog/2024/08/15...

https://www.ispreview.co.uk/index.php/2024/08/isp-ze...

In reply to a post by jpm:

I'd be amazed if someone was building an XGS-PON network on a 1:20 split ratio

Said 'contention ratio' rather than split ratio and a Google of that returns a result. Obviously he's confusing them, doesn't realise they aren't the same thing. Openreach would have to sell 2.4 Gbit/s connections to every customer, nothing lower, and to sell every port on the split including the two they don't generally use to reach the claimed 32:1 ratio.

Zzoomm sell up to 2 Gbit with most customers on a gigabit or lower. They could run a 1:128 split with the average customer on gigabit and be within the mentioned 20:1 contention ratio on the full fibre network. I note their use of 'on the full fibre network' too: the backhaul is likely way higher contention and it doesn't matter. A thousand gigabit customers will happily fit into 10 Gbit of backhaul with no visible slow down nearly all the time despite the 100:1 contention ratio.

We are in a Trooli area, and had flawless 1Gb connection 24/7 for two years.

Since we recontracted a few months ago to their 1gb/1gb service, the 'quality' is very poor.

Pages slow to load, and certainly a noticeable drop off in response at peak times, but their customer service is diabolical when trying to get hold of anyone.

If you eventually get through, you can talk to one person, then a. few days later when you call again and ask for the same person, they have left the company, and this has happened 3 times now.

You get no case number so have to go through the issues over and over again when you speak to someone new'.

Cant wait to bin them off as by the time we need to renew, maybe Zen might hjave better service.

Tried different DNS?

Have you been moved to CGNAT as a result of signing a new contract?

In reply to a post by Zadeks:

Tried different DNS?

Funny you should say that, as if i try to use Google or Cloudfare DNS, when the router reboots, the log still shows it connects to their DNS servers, even though in the settings i have changed them,

I tried a Netgear router, but that isnt as detailed as the Zyxel, and all it say is something like 'connected to internet', so i cant tell if it is their end or not.

AS for CGNAT, im not sure how i could find out? The 'external' ip address changes on each reboot.

In reply to a post by Alucidnation:

AS for CGNAT, im not sure how i could find out? The 'external' ip address changes on each reboot.

The IP address you get will be a good indicator. Just the first 2 of 4 would do, although if your IP address changes, obscurity does not really matter. Even public IP addresses can change with many ISPs

In reply to a post by Alucidnation:

AS for CGNAT, im not sure how i could find out? The 'external' ip address changes on each reboot.

Compare the IP shown on a site such as ipchicken.com or ipquail.com with the External IP shown on your router.

In reply to a post by jchamier:

In reply to a post by Alucidnation:

AS for CGNAT, im not sure how i could find out? The 'external' ip address changes on each reboot.

Compare the IP shown on a site such as ipchicken.com or ipquail.com with the External IP shown on your router.

Ok, so both of those websites results matches the WAN IP address on the router.


Is there a website where i can check what DNS servers i am connected to?

Edited by Alucidnation (Mon 16-Sep-24 17:02:04)

Then you aren’t on CGNAT….99% sure.

In reply to a post by jchamier:

Then you aren’t on CGNAT….99% sure.

In theory, if the WAN IP is the same as your public IP you could be on MAP-T, which is what Sky use for sharing the same IP address between 8 users. But the router needs to be using an ISP-assigned block of TCP/UDP ports for NAT, and this normally means you're using the ISP-provided router.

MAP-T isn't technically CGN, because the provider's network is stateless (that is, there is no stonking great NAT box that all the customer traffic goes through; the NAT state is held entirely on the customer routers). But it *is* a form of address sharing.

In reply to a post by candlerb:

MAP-T isn't technically CGN, because the provider's network is stateless (that is, there is no stonking great NAT box that all the customer traffic goes through; the NAT state is held entirely on the customer routers). But it *is* a form of address sharing.

Ugh! Sounds actually worse than CGN as you can't self identify I know that Sky also provide public IPv6, so at least not all is lost.

To be fair to Sky, their systems will automatically opt-you out of MAP-T if they detect you are using any application which won't work with it, and you can manually request an opt-out just by calling them up.

Hence it's completely transparent for the vast majority of users, and doesn't discriminate against people who want their own IP address.

They are extremely open, and gave a very detailed technical presentation at NetUK1. Slide 18 discusses opt-out, and slide 33 lists some third-party routers which are compatible.

Edited by candlerb (Tue 17-Sep-24 15:06:31)

In reply to a post by candlerb:

To be fair to Sky, their systems will automatically opt-you out of MAP-T if they detect you are using any application which won't work with it, and you can manually request an opt-out just by calling them up.

That is good, they must have nearly enough IPv4 then. As an older UK ISP that makes sense. Unlike the newer ISPs and alternate networks whom just couldn't get remotely enough.

They are extremely open, and gave a very detailed technical presentation at NetUK1. Slide 18 discusses opt-out, and slide 33 lists some third-party routers which are compatible.

Very cool, I've seen EE (BT) slides in the past on how they deployed v6 and run legacy v4 over it for mobile connections. (they may be the only UK mobile telco with v6 widely deployed).

I do like that title. IPv4aaS ... brilliant!

Edited by jchamier (Tue 17-Sep-24 15:55:28)

In reply to a post by candlerb:

In reply to a post by jchamier:

Then you aren’t on CGNAT….99% sure.

In theory, if the WAN IP is the same as your public IP you could be on MAP-T, which is what Sky use for sharing the same IP address between 8 users. But the router needs to be using an ISP-assigned block of TCP/UDP ports for NAT, and this normally means you're using the ISP-provided router.

MAP-T isn't technically CGN, because the provider's network is stateless (that is, there is no stonking great NAT box that all the customer traffic goes through; the NAT state is held entirely on the customer routers). But it *is* a form of address sharing.

Interesting, thanks.

Although not sure i completely understand.

I have used my own Netgear router on this connection and that worked ok?