Upgrading the NBN with G.fast has its limitations

Quite coincidentally, at the same time that G.fast is being discussed in Australia a similar discussion is taking place in the USA; and there is doubt there too about the contribution that G.fast can make to improve the performance of the faltering broadband systems in both countries.

G.fast is a band-aid solution that can be bolted onto the VDSL/FttN technology, but it only works at very short distances, and even then there are plenty of technical problems that will stand in the way of G.fast providing a positive contribution to the NBN beyond some niche market situations.

As we predicted exactly a year ago, aware of the desperation of the nbn company, vendors are only too willing to come up with ‘fix-up’ solutions in an attempt to sell technology to the broadband company.

In that blog from August last year there is an extensive overview and explanation of the G.fast technology in relation to the NBN, as well as an in-depth overview of the issues associated with the technology.

You might recall my American colleague Fred Goldstein whom I have quoted regularly in my blogs. This is his latest comment on the G.fast technology:

In all reality G.fast has little to do with fixed landlines-based broadband infrastructure, since it’s only a short-hop mostly-indoor distribution technology. Think of it as sending GigE down crappy old Cat3 wire instead of having to pull clean Cat6 or 5E.

 It only works for a few hundred feet, but if a building’s interior phone wire (which is normally owned by the building, not the telco) is not too crusty, then this will let a very fast fiber-connected DSLAM in the basement feed the apartments upstairs with faster speeds than an outside-plant DSL could supply.

 Its main advantage is not having to pull fresh fiber to each apartment, noting that fiber has bend radius issues (the newest variants keep getting better at this, though) which don’t apply to copper, and which can be problematic when snaking down hallways.

Experts in other countries are also questioning G.fast – in particular the effect the deployment of this technology will have on other DSL services that are provided within the same infrastructure system.

We have also reported on netBlazr. This US company is in the process of installing similar equipment but based on G.hn. As with other VDSL, they use it for the last 30-200 meters, that is, entirely within a building that has only older phone wire and no easy way to add Cat5e or Cat6 cable.

Whereas G.fast assumes one vendor controls all the pairs in a cable, G.hn assumes there will be other DSL technologies in use in the same cable.  Given that all their links are extremely short it is claimed they should see similar performance (up to 800 Mb/s down and a few hundred up). However if they can get >500 Mb/s down and something plausible up, they will have killer upgrade for legacy buildings, where previously netBlazr was only able to get Fast Ethernet speeds (100/100 Mb/s) using VDSL profile 30a.

So it is clear that the discussion on G.Fast is far more complex than a simple bolt-on and it will do little to improve the overall performance of the NBN.

Paul Budde