In 2016, G.fast looked very promising.
Thousands worked at developing and deploying.
It wasn't enough.
Most carriers are investing
in fiber or 5G instead.
Dark Blue: Firm commitments from incumbent: BT (10M), Belgacom, Australian NBN, Swisscom, Austria, Bezeq Israel, Chunghwa Taiwan, Telus Canada, Telekom South Africa, SK Korea, (U.S.) AT&T, Century, Frontier, Windstream, Belgium, Omantel
Mid Blue: Smaller carriers in Germany, Norway, Finland, Japan
- Published: 23 September 2016 23 September 2016
Pioneers get arrows in their back and schedules do slip. G.fast technology is rapidly improving, with Amendment 2 and Amendment 3 chips delayed until the second half of 2017. BT had originally promised to start their 10M home rollout late this year or early next. In particular, BT wants 48 or 96 ports while today's tech only supports 16-24 ports. They also want the higher power and longer reach of the Amendment 2/3 chips. That's particularly important. Their last goal (350 megabits down at 350 meters) is beyond the current state of the art. It may be impossible, but that's unclear.
The original plan, promised by CEO Gavin Patterson, would go to local distribution points for speeds generally 500+ megabits. The 500 meg promise was part of Gavin's attempt to persuade OFCOM not to break up BT because they would deliver great Internet to "nearly all of Britain by 2025." Then the bean counters jumped in and said they wouldn't pay for the high speed. Using the existing cabinets brings the cost so low it fits into the existing capex budget but also reduces the speed.
Some very good BT engineers redesigned the project for 300 meters rather than 100 meters.
They brought the changes (more power, more bits/signal) to the ITU standards committee. Broadcom and Sckipio, the main chipmakers, went to work. They tell me it's been challenging, especially with the heat and power budget. In addition, vectoring 48-96 ports requires massive processing power, adding heat. Adtran's solution, which works well for VDSL, is to move the vectoring engine into an external module, the SLV. That's inspired a debate about whether vectoring need to be onboard or can be external. The Adtran SLV isn't ready for G.fast in any case.
I'm going to call the downstream speed 200+ megabits, rather than the "up to 330" in the BT pr. It's time to use actual speeds that most users can achieve. Anything else is false advertising. BT's regulator, OFCOM, has promised to clamp down on false claims. So have the FCC in the U.S. and BnetzA in Germany. So far, they've all been afraid to take on the telcos. Deutsche Telekom has been particularly egregious, building a vectored VDSL network designed for 30-50 megabits to many as "up to 100." Frontier isn't even vectoring, but also claiming "up to 100."
They are expanding the trials to 140,000 homes. Light Reading expects that by March. If BT concentrated on apartment buildings, the 500 meg is easily achieved already. But Brits love their gardens, and only about 11% live in MDUs.
BT's speeds can be increased by an important 100 megabits by eliminating the VDSL carveout.