In reply to a post by WilliamGrimsley:

Yes, that's what I meant by national.

I'm no wiser, sorry if I'm being obtuse.

Edited by deleted (Fri 17-Feb-17 00:12:12)

In reply to a post by WilliamGrimsley:

Really? I'm not aware of any national rollout.

I never said there was any rollout. I did specify BDUK areas. Pretty happy with how I worded it.

OpenReach will likely never rollout vectoring to the majority of cabinets. There is still lots of cabinets using it, with nothing to do with any trial.

In reply to a post by j0hn83:

OpenReach will likely never rollout vectoring to the majority of cabinets.

I did start to wonder whether it would happen through natural attrition of cards in DSLAMs. Through Moore's law, vectoring becomes "easier" in silicon, which *might* have made for an easier deployment eventually.

However, it looks like BT might be choosing to upgrade cards to ones with more ports (48-port cards to 64, and 32-port cards to 48). When I see the specs for cards that have higher port counts, they seem to conspicuously not support vectoring.

The improvements in silicon seem to be being directed at the extra port count, rather than to vectoring.

In reply to a post by WWWombat:

I did start to wonder whether it would happen through natural attrition of cards in DSLAMs. Through Moore's law, vectoring becomes "easier" in silicon, which *might* have made for an easier deployment eventually.

However, it looks like BT might be choosing to upgrade cards to ones with more ports (48-port cards to 64, and 32-port cards to 48). When I see the specs for cards that have higher port counts, they seem to conspicuously not support vectoring.

The improvements in silicon seem to be being directed at the extra port count, rather than to vectoring.

The Huawei kit is vectoring ready, with addition of a vectoring engine board. The ECI kit doesn't have a fast enough backplane to support vectoring.

What did you have in mind for line card silicon to accomplish?

I dunno ... I'm not a hardware engineer.

I just know that, for example, the Huawei 48 port cards do have vectoring-ready variants that could be used in the 288 cab. But the newer 64 port cards that would turn it into a 384 cab don't show a vectoring capability.

I would presume there is a physical limitation being hit... Not enough space on the linecard, not enough power capacity, insufficient cooling capacity perhaps. All of which, to me, are rooted in trying to ask too much of the current generation of silicon they can create.

Or it could be financial - no one will pay.

Or just timescales - they haven't had time to create that variant yet.

Wonder if it's nothing more than that the line card connection to the backplane isn't fast enough to handle vectoring data for 64 VDSL 2 lines, or the VEB doesn't have a fast enough connection to the backplane to handle 384 VDSL 2 lines.

A few years ago Huawei informed that 20Gbps per slot of connectivity to the VEB was needed. If the backplane is 120Gbps per slot and vectoring 6 x 64 port cards needs more that'd explain it.

Just speculation. As you suggested the actual silicon on cards should be fine to process the cancellation data.

As you say, also, processing wise a newer VEB should be capable.