Speeds are fine, "Unless there's a line problem." I've been reporting for three years that ~10% of lines have problems. In the chart by Rami Verbin of Sckipio, he finds G.fast goes ~130 meters on good lines. Poor lines have about half the reach. His chart roughly matches the reports from Swisscom, Belgacom, and England for both G.fast & vectored DSL.
The causes include bridge taps, paper insulation, and landlord-installed cheap wiring. The 10% with problems can cause the majority of the line-related complaints to support. The angry customers drive up cost.
Rami's solution is bonding, supported on the Sckipio chips.
Bonding has been part of the DSL standard for more than a decade. Sonic.net in California has made it work for two lines. Several companies have bonded 6-10 lines for a robust commercial service.
Large telcos have generally avoided bonding. Besides the actual time involved in going to the field cabinet, the companies don't want to carry additional parts or train their army of installers. AT&T has for several years been officially offering a bonded service in order to claim more competitive speeds. I believe they effectively discourage customers from ordering the higher speeds and have installed relatively few in the field.
Verbin made some additional points:
- 4 gigabits is possible by bonding two decent 2 gigabit lines.
- Even in a service from remote cabinets, ~25% are close enough to get a full gigabit. BT has chosen to limit their G.fast to 160 or 330 meg and throttles everyone else down. AT&T did similar in ADSL days. They believed customers would complain if some got 20 meg and others were much slower. I always thought they were making a mistake.
- cDTA and iDTA are practical ways to deliver much higher upstream by switching some bandwidth from downstream to upstream only when needed. Sckipio presented the idea in standards in 2012 and report it is now working well.
35B will probably be similar but Deutsche Telecom doesn't expect to deploy until 2019.