Thanks @chrys
I've watched that video on a number of occasions now, as it was one of the first that gave us proper ideas of what BT were up to for both G.Fast and connectorised fibre. Its a shame they haven't kept technical communication quite as open since then.
PPPoE:
Yes, Neil instigated the mention of PPPoE, but only as far as a nice-to-have, "on the list of things to solve". When I said "no indication", I really meant since then; that there was no indication in the current service-defining documents. No matter what Neil said back then, I'd expect reality to be reflected here and now - within STIN 520 on one side, but in a WBC document on the other.
STIN 520 does nothing to indicate any change of focus away from PPPoE. Even when it has a section warning CPs of the long-term possibility of needing battery backup (SOGFAST?)
I did find
this tech brief from Cerberus that tells us PPPoE is used in the router still.
Perhaps, in the end, PPPoE is rescued by hardware acceleration. Here's a table of
TP-Link router throughput, where there seem to be a few capable devices. I wonder what this will do for the number of MSE bRASs required.
It is also Neil himself who said the existing capacity to the cabinet is 2.5gbit. I take his answer to the question not been evasive, it came across to me as, he expects issues when reasonable take up occurs, but its not a problem because things are in place to add capacity as and when it is needed up to a maximum of 40 x 10gbit..
I think everything in your second sentence is right, but not the first.
The questioner mostly asks about cabinet backhaul generically, and mentions that it is not a known quantity to LLU CPs (so presumably he's really talking about FTTC backhaul). He does briefly state that he thinks they are serviced in shared PONs, but mostly stays generic with the question. Neil's answer isn't evasive, but just focusses on the generic needs of a G.Fast node. I think.
(Context: The video pre-dates the announcement of the G.Fast pods, so mention of backhaul to the cabinet is probably based on an assumption that G.Fast will be in/next to the FTTC cab)
I do get the impression (but only an impression) that G.Fast backhaul will be PON-based. The impression comes more from wider industry behaviour than from specific BT information, which is otherwise lacking. That Neil's answer only talks about PON is perhaps a tell, though.
My suspicion, then, is that Neil was aware that G.Fast backhaul will be PON-based, and answered the questioner with a focus on G.Fast only, not FTTC. His answer is that GPON speeds will not be enough, but XGPON and NGPON2 will be enough.
My further suspicion relates to why G.Fast and Connectorised FTTPoD both came labelled "NGA2" in this presentation. This suspicion is based on the previous one (that the G.Fast nodes will be fed by PONs), so it will therefore kickstart the deployment of more splitter nodes (ie in FTTC areas). Such deployment of splitter nodes means that FTTPoD no longer needs to go all the way back to the Aggregation Node - instead, being worked only to the Splitter Node ... and hopefully only charged that far.
In which case ... if FTTPoD is going to get cheaper, I bet it does so with the proviso of being in a G.Fast area.
There's a lot of guesswork here, though, and little concrete fact.