G.fast nationwide deployment in 2016-7 - pilots in Huntingdon & Gosforth in 2015

[#1]

Quote from: chrcoluk

except g.fast requires vectoring to work properly.  So presumably if vectoring doesnt work then there is little point with g.fast.

And G.fast requires entirely different hardware, not least different DSP capabilities, different ways of being powered, a different number of lines supported, and a different way of being packaged. So presumably the status of vectoring on the ECI VDSL2 cabinets that are already standing has little impact on anything whatsoever.

[#2]

Quote from: Melancholie

Regardless their DSLAMs can be used to backhaul G.fast nodes in the field.

Is that true? In an integrated way?
As far as I can tell, Alcatel-Lucent treat their 7357 ISAM nodes as a remote extension of the main cabinet, and all O&M works through the main cabinet.
If an intermediate cabinet is just going to be part of the backhaul by acting as a glorified splice, isn't it better to wire the G.fast node directly into the aggregation node?

[#3]

Quote from: Melancholie

Quote from: AndyH I doubt ECI have anything like the capital for R&D as Alcatel/Huawei. Indeed. I'm sure they'll have G.fast line cards ready at some point they'll just be a bit behind the curve.

Huawei's mission statement appears to become like Ericsson. That'll need a lot of R&D money.
Broadband Trends had this image for the relative maturity of the vendors for vectoring in 2012:

[#4]

Quote from: WWWombat

If an intermediate cabinet is just going to be part of the backhaul by acting as a glorified splice, isn't it better to wire the G.fast node directly into the aggregation node?

You wouldn't use it as a splice, you'd use it as a switch. The MA5603T can take 16 port GPON cards.

[#5]

Quote from: WWWombat

Huawei's mission statement appears to become like Ericsson. That'll need a lot of R&D money.

Surely you mean ALU (Alcatel) rather than Ericsson?

[#6]

Quote from: Melancholie

You wouldn't use it as a splice, you'd use it as a switch. The MA5603T can take 16 port GPON cards.

You could, but why?
The point of the AGN is to splice the fibre to nodes or users further out. There's no need to force everything the FTTC cabinet - especially if that forces less capacity for users within range of that cabinet.

[#7]

Quote from: AndyH

Surely you mean ALU (Alcatel) rather than Ericsson?

Nope, I do mean Ericsson. Speaking as someone who worked for Ericsson, and knows people who went to work in Shanghai... I've heard precisely just how much they wish to emulate Ericsson
Ericsson has a much greater international presence, particularly in mobile - and I mean the network side, rather than the device side. They're very big in telecoms - but the public knows almost nothing about them, save their little side-trip into devices in the eighties and nineties.
Edit: I was trying to say that Huawei were trying to emulate Ericsson across everything they do, rather than just in the narrow field of vectoring; in that field, obviously ALU would be a more natural target.

[#8]

Don't get your hope up too high because G.fast likely to be hit and miss with speed via distance from cabinet to your property below:
    500Mbps @ 100m
    200Mbps @ 200m
    150Mbps @ 250m

[#9]

The distance, of course, depends on the location that BT choose to put the node. If it is placed where the DP is currently placed, then the distance will be the same as the existing lead-in wire or drop wire.
As part of the trial, BT pointed out that two of the distances used (the short line of 19m, long line of 67m) represented the 20th and 80th percentiles of the distribution of drop-wire lengths in the country.

[#10]

Quote from: WWWombat

You could, but why? The point of the AGN is to splice the fibre to nodes or users further out. There's no need to force everything the FTTC cabinet - especially if that forces less capacity for users within range of that cabinet.

Can't put many G.fast nodes onto a GPON split. Saves blowing further fibre from AN to each area - there isn't an abundance of it blown initially.
Doing it the other way means basically doing an FTTPoD build for each node. Quite time consuming if deploying to distribution points.
The cabinets are there - finding other uses for them when VDSL is obsoleted makes sense.
Shouldn't affect capacity for VDSL users directly connected to cabinets.