G.fast: Niche or Not?
At last week’s Broadband World Forum, G.fast, the technology of choice for FTT Distribution Point (FTTdP) deployments, dominated the debate about the evolution of fixed access networks, showing just how far and how quickly it has come since first mooted in 2011.
Yet it still remains unclear exactly how far it will go, and where it really makes sense; with operators increasingly seeking to hitch themselves to the Gigabit bandwagon, the lure of an all-fiber solution remains strong for many.
Operators and vendors looking at G.fast have no doubt been encouraged by the recent success of VDSL vectoring, which has gone from near-zero to 15 million lines delivered in only two years. Vectoring has shown that there are still plenty of telcos, especially in Western Europe and North America, that want to retain some copper in the access network for as long as feasible.
Could G.fast achieve similar success? Certainly it is attracting more attention than many had anticipated back in 2011. Then, there was a strong sense that G.fast might be too late to make a real difference, and would at best be a short-term niche solution. But with the standard set to be formally ratified in December, several dozen operators testing it in lab and field, and with leading vendors readying product launches, we could be seeing the first deployments by the first half of 2016.
Yet despite the high level of interest, there are still some important questions to answer. Not least, initial products, operating up to 106MHz, will not enable operators to market "gigabit" services, since the maximum bandwidth available in a typical deployment scenario (50-meter loop length and multiple users) will likely be around 700Mbit/s (divided between upstream and downstream). A proposed G.fast variant uses 212MHz, and will potentially supply more than 1 Gbit/s per customer, but it's not expected to be commercially available until 2017.
Nor is that the only issue. Among others raised at BBWF, the lack of a fully baked operations and management environment is a significant barrier, alongside related issues such as vendor-to-vendor interoperability. Many operators also expect reverse power feed. As BT noted, the operations issues are non-trivial: G.fast will likely entail supporting 20 times as many nodes as VDSL, typically mounted on a telephone pole supplying up to 8 homes.
Many of these issues are being addressed by the Broadband Forum (BBF) in its (unratified) WT 301 and WT 318; the BBF's certification program will increase operator confidence, but likely delay decision-making for now.
For vendors, meanwhile, the need to support a range of deployment scenarios is a further headache, with the number of options (the BBF counts 23) including deployment on poles, in manholes, in MDU (multi-dwelling unit) basements and apartment floors. Yet another issue is the co-existence of G.fast alongside VDSL, which is required by many operators; although this is built into the standard (and has been successfully demonstrated in several field trials), it's expected to reduce the bit rate by 20% or so.
Alongside these technical and operational issues, telcos must determine what the business case is for G.fast, and whether a further interim step before deployment of FTTH is really justified. While the capex differential between FTTH and VDSL is at least 3 to 1, it's more like 1.3 to 1 for FTTH to G.fast. Add in opex considerations, and the cost-savings over FTTH, at least as a mass-market solution, look underwhelming.
In a cautiously worded presentation, Trevor Linney of BT (a strong promoter of work on G.fast) told delegates that it was very interested in the technology, and had run several successful field trials, yet was not yet ready to commit to use it -- not least because it's not yet sure when it will need a product that delivers hundreds of megabits per customer.
G.fast, in other words, is a very different entity from VDSL vectoring. In the fragmented fixed access market, there are many national markets where G.fast most likely will find very little interest. But in others, despite the barriers described above, G.fast could yet play a significant role. In particular, it could be a very useful solution wherever it is difficult to get fiber inside a building, for instance, when connecting historic buildings or where landlords are particularly intransigent.
Equally, G.fast may be useful where it's too difficult to replace the twisted pair internal wiring of MDUs with optical fiber or Ethernet cable. In these scenarios, the cost advantages of G.fast may prove decisive. And it has some handy features such as the ability to vary upstream and downstream ratio to suit different customer segment requirements; that could work in some mixed deployment scenarios.
G.fast, then, is no panacea, but it's a useful tool in the telco battle to upgrade the fixed access network for the 21st century.
— Graham Finnie, Chief Analyst, Heavy Reading