I don’t think the “split-less” distance is as long as 70km for GPON. Huawei maintain the maximum physical reach is 40km for class B+ and 60 km for class C+ optics. I believe Openreach optics are class B+ (at least my ONT is). Of course these distances are maximum theoretical; splits, splices and connectors all take their toll on the budget.

There is also a maximum differential distance between ONT/Us of 20km.

https://info.support.huawei.com/AccessInfoTool/PON_B...

Edited by Pheasant (Fri 12-Mar-21 17:24:00)

In reply to a post by jabuzzard:

... though at that point you are probably better off using some Ethernet BiDi SFP's than messing about with OLT's which are a lot more expensive per port than an ethernet switch.

On a single link basis, yes, but on broadband deployment, no. Remembering a single GPON port is effectively shared / multipoint, the “cost” per user is a fractional basis. Basic chassis cost for OLT is about the same as an decent L3 switch. A 2-slot Huawei MA5800-X2 OLT with management cards, 10G uplinks + dual PS is less than $3K. ONUs are really dirt cheap. Single port Huawei ONT are about £15 wholesale

At the other end of the scale an 8 no. GPON port Ubiquti OLT is around £950. Their own ONUs are something like £25 to £45 each.

Edited by Pheasant (Fri 12-Mar-21 18:36:06)

In reply to a post by Rich44:

Just wondering what parts of the Openreach fttp kit actually need power, I realise most of it is passive obviously but what bits likely to be in the street, Chambers or up poles needs power?

On your end as a customer, the ONT.

Nothing in the street, nothing in chambers, nothing on poles. Hence "PON" - Passive Optical Network.

In reply to a post by Pheasant:

In reply to a post by jabuzzard:

... though at that point you are probably better off using some Ethernet BiDi SFP's than messing about with OLT's which are a lot more expensive per port than an ethernet switch.

On a single link basis, yes, but on broadband deployment, no.

Well a number of broadband fibre providers would disagree with you. For example B4RN is all PtP BiDi, as is Wightfibre. The cost of the PtP solution has come down significantly faster than GPON. It is also an utterly commodity item so if you want to switch vendors well not remotely a problem.

However the context of the post was that in various places Openreach are going to much lower splits (1:4) with subtended OLT's to get better reach in the "deep rural" areas which is probably reserved for the remotest parts of Scotland. My comment is at that point BiDi ethernet where 120km SFP's are a commodity item is IMHO a better solution. SFP ports are as cheap as chips these days, and the significantly greater reach removes all that out in the field powered equipment the subtended OLT's will require.

At the end of the day GPON looked like a good idea even 10 years ago. However as always in IT things change and if you where starting again today from scratch simple point to point BiDi ethernet looks a better bet.

You have zero vendor lock in, much easier migration paths for higher speeds; for example you could offer symmetric 25Gbps speeds today and igher flexibility in the physical network.

Edited by jabuzzard (Tue 16-Mar-21 22:23:29)

In reply to a post by Pheasant:

I don’t think the “split-less” distance is as long as 70km for GPON. Huawei maintain the maximum physical reach is 40km for class B+ and 60 km for class C+ optics. I believe Openreach optics are class B+ (at least my ONT is). Of course these distances are maximum theoretical; splits, splices and connectors all take their toll on the budget.

There is also a maximum differential distance between ONT/Us of 20km.

https://info.support.huawei.com/AccessInfoTool/PON_B...

I was using a Nokia formula for your optical budget based on number of splits and length of fibre. If you run the numbers for class C+ you get to 70km when you have no loss because you have no splits. It's the splits that really introduce the loss. I am assuming in that a PtP link with no splices. So pearing that back to 60km would probably be prudent, as you are going to need some splices outside the 70km of fibre sitting on a drum in a lab.

Given you can change it on a per port basis on the OLT by plugging in a different SFP nobody is going to get a class C unless they need it.

Going beyond discussing bidi vs pon optics for a moment....

I think they’re quite interesting examples that you cite. Both these nets have somewhat unique ‘environments’ in which they have built their networks. B4RN have a relatively small yet devoted user base of largely rural folks that either contribute their own labour and/or wayleave-free in-verge or land to the cause. Wightfibre similarly quite a rural place - therefore I would argue these aren’t your typical mass scale deployments. I would call these “green field” deployments, where cost per metre is generally quite low and there is more opportunity to have significantly higher fibre count in the ground with less penalty.

The design flexibility of PON is one of it’s biggest draws. The fact is that far less fibre and splicing is required for PON. This a critical consideration especially in dense urban or “brown field” deployments, where high fibre count means high ducts requirements. To try and offset this with PtP requires active hardware to be placed deeper into the network, closer to subscribers. Goes without saying that this has consequences not only for higher numbers of active devices, but also ongoing operational costs to maintain and provide power etc.

The design flexibility of PON makes it a good candidate for dovetailing into existing physical infrastructure, like Openreach PIA. Trying to “do” PtP if your wanting to leverage PIA instead of digging your own ducts from scratch bound be far more difficult proposition. I think this is why networks that are deploying in brown field are opting towards PON architectures.

The other aspect to consider is the so called “future proof” nature of PtP. I don’t think one could argue for that perspective as it’s pretty hard to deny that a dedicated strand(s) of glass to each subscriber is the gold standard of a fibre network. However I would say that technology moves fast and the familiar GPON standard has already been skipped by several forward looking alt nets who have adopted XGSPON which addresses bandwidth and symmetry.

Feels like a rerun of Betamax versus VHS!

Just a quick summary (not exhaustive) of some network providers currently installing and/or running PON type nets (many XGSPON) in the UK:

- Openreach
- CityFibre (incl. Air Broadband, Giganet, Highnet, TalkTalk, Triangle Networks, Trunk Networks, Vodafone, and Zen Internet)
- Connect Fibre
- Community Fibre
- Exascale
- G.Network
- Glide
- Hey! Fibre
- Hyperoptic
- Swish Fibre
- Telcom
- toob
- Trooli
- Zzoomn

I’m aware Gigaclear (again mainly rural), but what others are running PtP?

Many years ago, I worked for a large ISP with thousands of dialup modems. Originally they had chassis-mounted analogue blades - there was a separate incoming wire for each phone line. Then they changed to E1 trunks, so one cable carried 30 channels. The saving in both rack space and cable spaghetti was enormous.

The same applies to PtP versus PON. Although you can get switches with high density SFP ports, and more dense than PON blades, you would still need around 30 times as many incoming fibres to service the same customer base - and of course, 30 times as many fibre strands running to your fibre aggregation nodes.

Hence for mass-market consumer FTTP, I think PON still makes a huge amount of sense - and clearly most of the providers do too.

Certainly there will be business users for whom this is not suitable, but that's a much smaller problem to get dedicated fibres to those customers.

In reply to a post by Pheasant:

The design flexibility of PON makes it a good candidate for dovetailing into existing physical infrastructure, like Openreach PIA. Trying to “do” PtP if your wanting to leverage PIA instead of digging your own ducts from scratch bound be far more difficult proposition. I think this is why networks that are deploying in brown field are opting towards PON architectures.

If you can draw a single core of fibre to a splitter you can draw 32 cores, and the cost is almost all in the man power. There is indeed less splicing as any splicing in segments before the splitters is reduced. Once you get to the splitter the amount of splicing/termination is actually more with PON. However again most of the cost is in the labour for pulling of the fibre.

The other aspect to consider is the so called “future proof” nature of PtP. I don’t think one could argue for that perspective as it’s pretty hard to deny that a dedicated strand(s) of glass to each subscriber is the gold standard of a fibre network. However I would say that technology moves fast and the familiar GPON standard has already been skipped by several forward looking alt nets who have adopted XGSPON which addresses bandwidth and symmetry.

Feels like a rerun of Betamax versus VHS!

Even XGSPON does not compete with PtP BiDi ethernet. Where is my symmetric 25Gbps connection today?

The use of PON technologies is historical. Things have moved on and the assumptions that favoured PON have mostly evaportated. If you where starting again PtP ethernet has lots going for it. It is a bit more expensive, but it is vastly superior in the long term.