
Summary Bullets:
• G.fast, ratified a week ago by the ITU, can deliver up to 1 Gbps of aggregate throughput to end users over the existing telco copper plant. With testing and some limited commercial deployments planned for 2015, is the hype about G.fast deserved?
• Will developing cable access technologies such as DOCSIS 3.1 and CCAP provide cable operators with effective countermeasures to G.fast and other telco copper-focused enhancement technologies?
One week ago today, on December 5th, the ITU ratified the G.fast standard. For many industry insiders, this was no surprise. In fact, it was simply the formal approval of G.fast specification G.9701, per the ITU’s schedule. However, the media buzz generated by the ITU’s G.fast approval has seemed disproportionately higher than expected. So, what gives? Is December a slow news month? Yes. Are there not more important things to write about in terms of fixed broadband? Actually, no. Finally, does G.fast deserve all of this attention? Well, yes!
We first heard rumblings about what would become G.fast in 2011. At the 2011 Broadband World Forum, Huawei mentioned development of a multiple input multiple output (MIMO) and Super-MIMO-based prototype for ultra-high-speed bandwidth over copper, supporting speeds well beyond those delivered by VDSL2 vectoring. Several months later (December 2011), Huawei announced its “Giga DSL” prototype, and for the first time, the ability to deliver 1 Gbps connections over 100 meters of telco copper, although even then, the Gigabit was aggregate (downstream plus upstream) bandwidth. Until that point, Gigabit service delivery over anything but FTTH was a pipe dream.
Fast forward three years, and here we are with the G.fast standard ratified and ready for prime time. What has emerged from the comprehensive standards-development process – with massive input from telecom equipment vendors such as Alcatel-Lucent, Huawei, and ADTRAN, silicon suppliers, and telco operators – is fiber-to-the-distribution-point (FTTdp) technology. And despite the small letters, the “fast” portion of G.fast is an acronym for fast access (to the) subscriber terminal.
The distribution point is simply the node in the network where the (network) fiber is terminated, and then connected to the subscriber-facing last-run copper (ideally <100 meters). The distribution point unit (DPU) is the network hardware node performing this function. The first generation of DPUs are available as single-port (in cases where fiber is deployed as a de facto FTTH solution for single-family home applications), or multi-port (with eight, 16 or more G.fast subscriber ports serving closely-clustered residences, including apartment buildings).
So, what does the advent of G.fast really mean? Will G.fast bring higher speeds to more consumers, or just a select few? Finally, if G.fast means taking fiber almost all the way to the end user, why not just go full fiber (to the home)?
G.fast once again extends the life of telco copper, or at least the most-expensive-to-upgrade portion of the telco copper plant (that closest to the end user). While VDSL2, vectoring and bonding technologies also enable telcos to deliver higher speeds – even “fiber” speeds – and expand their serving areas, G.fast’s ability to support 1 Gbps is a game-changer.
How? G.fast gives telcos the ability to compete very effectively over the long term with cable operators, which have collectively and cooperatively (via CableLabs) been developing an array of technologies designed to compete with FTTH (including DOCSIS 3.1, CCAP) while still leveraging their hybrid fiber-coax (HFC) networks. DOCSIS 3.1 will deliver 10 Gbps per node downstream, giving cable operators a strong counter-punch to existing VDSL2/vectoring-based offerings from telcos. The end result is that through true intermodal competition (telco vs. cable), consumers will benefit as rival operators escalate their connectivity speeds, with G.fast and its Gigabit capability enabling telcos to likewise counter DOCSIS 3.1-based cable offerings.
And what about the FTTH option? With G.fast, telcos will still be required to make an investment in “deep fiber” – but FTTdp/G.fast deployments are projected to be a lot less expensive than full-fiber FTTH network builds. G.fast chipset vendor Sckipio asserts that G.fast is up to US $1,200 cheaper than FTTH per household (primarily due to labor/trenching costs for connecting fiber in the node-to-household portion of the network). While a $1,200 differential needs to be viewed with some, ah, sck-epticism – as it bolsters Sckipio’s business case – the reality is that even at half that cost, the savings is massive, especially when the telco can deliver the same “fiber speed” connection to its subscriber base.