NBN zealotry in the ultra-high definition age
Do anti-FTTP NBN arguments stack up against the reality of UHDTV and Terabit Ethernet?
Australia’s Shadow Communications Minister Malcolm Turnbull recently declared the IT media includes a number of “zealots” who won’t, such is their/my fanaticism, report fairly on his alternative National Broadband Network (NBN) plans.
One of his chief beefs was that the press corps are collectively ignoring the fact that fibre-to-the-node, (FTTN) Turnbull’s preferred NBN alternative, works rather well delivering a triple play service in other parts of the world.
Turnbull’s argument for FTTN stands on three principles.
The first is that it will be cheaper to build than the current fibre-to-the-premises (FTTP) plan. The second is that it will be faster to build, which is a good thing because the faster businesses and homes get their hands on broadband the better it will be for Australia's economy.
Lastly, he argues FTTP is too good, and that most homes and businesses don't and won't need all the bandwidth it can provide. The third point is the foundation for the first two, as it posits FTTP as wasteful over-investment that goes beyond what is necessary.
If only the press corps would look overseas at FTTN and wireless trends, Turnbull argues, we'd see that these arguments stack up and realise the colossal price tag for the FTTP NBN is
stealing the future from our children needlessly extravagant.
Turnbull's right about one thing: your correspondent has never considered overseas FTTN plans.
But The Register has, just a few days ago, noted that the IEEE has started work on Terabit Ethernet. And not just Terabit Ethernet – it’s also planning Ten Terabit Ethernet, because it sees demand for faster networks and is moving to stay ahead of the market.
We’re pretty sure Turnbull hasn’t considered those overseas developments and the powerful signal they send about likely future increases in demand for network capacity.
What does that mean for his three arguments?
Turnbull's point on network build costs is moot: the complications of NBNCo's arrangements with Telstra mean it is all-but-impossible to predict a final cost for an FTTN NBN. On time-to-build, Turnbull has a strong argument when one considers the hybrid fibre coax (HFC) networks built for cable television could have delivered more fast broadband, more quickly, than an FTTP build. It's hard to see just how FTTN on a national scale would be significantly faster to roll out than FTTP, especially when one considers the likely renegotiations required.
That adds up to a big stalemate on both of those arguments, especially absent a formal plan from Turnbull.
That leaves us with the argument of FTTP's suitability. FTTN and wireless are both touted as capable of delivering around 100 Mbits/s to each end user as soon as or faster than it will be possible to install near-universal FTTP. 100 Mbits/s delivers a fine internet experience in Australia today, which is probably why NBNCo reports more punters than it expected are buying such connections. Fibre can of course snack on 100 megabit connections and has already shown very strong potential to carry terabits of data per second.
That’s not us saying punters will queue for Terabit connections any time soon, by the way. Indeed, we mention Terabit Ethernet only because, during your correspondent's time in the industry, Gigabit Ethernet went from seeming impossibly capacious to being an unremarkable feature in home routers. The exotic routinely becomes … err … routine in this industry.
The NBN is planned to have a 30+ year life so it is not inconceivable that it could be asked to carry terabits per second before its successor is commissioned. The considerable likelihood that fibre can take on that task surely makes it a more worthy investment.
But that's still decades away, so we need to ask if FTTP is worth it in a shorter timeframe.
To do so, let's consider the dominant application for today's public networks, namely carriage of video. Today's connections can cope with high definition transmissions, but if you've spent more than an hour attempting to download an HD movie from iTunes you know it cannot do so well (or do so without Akamai and other such tricksters in the background helping things out). today's networks will almost certainly struggle to cope with UHDTV when it arrives in a few years' time given that David Wood, chairman of the ITU-R Working Party 6C that cooked up UHDTV has told The Reg that even the new HEVC compression technique will probably mean a single stream of 8K UHDTV will need 90Mbit/s. A single 4K stream will need 25Mbits/s.
Wood thinks there’s another compression standard coming that could halve that again. Bu that still leaves 8K UHDTV needing 45Mbit/s, which is nasty even if, as seems likely, the standard looks like being a curiosity for some time.
Let's be charitable and imagine that in 2020, HEVC's successor arrives just as 4K UHDTV-enabled monitors and TVs start to land in homes. Let's also imagine, as we recently reported, that IPTV takeup surges to 27% of homes by 2016. By 2020 we could generously post that 30% (about today's Foxtel takeup) of Australian punters would need at rock-steady 12 Mbits/s to enjoy a single IPTV 4K video stream.
ADSL2+ won't do that job. 12Mbit/s FTTP connections may struggle (as may the NBN's satellite connections, but that's another story). Various FTTN rigs promise they'll deliver the required speed , as do some wireless futures. Whether they'll do them well is harder to say, given the near-inevitability of a need for the same kind of overhead (and Akamai-esque helpful interference) to be built into tomorrow's networks as is commonly built into today's.
At this point in the argument it gets hard to wrap things up. A zealot would say that the advent of UHDTV and Terabit Ethernet point to a time when demand for bandwidth will rise to extraordinary levels and that as FTTP is the technology most likely to meet that demand it is therefore the most prudent choice. As someone who started online life in 1993 with a 2400 bits per second modem, and now luxuriates on a 14mbit/s ADSL2+ connection (when Telstra's pits are dry), I have sympathy for that argument.
Yet that ADSL2+ link is a challenge, too, because its five-year reign as Australia's dominant broadband standard has made for a plateau in innovation: service providers have only had ADSL2+-fuelled expectations to meet for five years. Absent faster connections, innovators have not had a reason to consider applications which demand more bandwidth.
Might they do so in future?
Re-enter, at this point, consumers' unexpectedly strong takeup for faster NBN services, a sign punters want speed.
Re-enter also UHDTV, the HEVC sucessor and Terabit Ethernet, complete with the powerful signal they send about future demands and needs.
None, to be fair, seem likely to put pressure on Turnbull's preferred FTTN NBN configurations by the time it is completed. But on the NBN's timeframes, it is foreseeable all will make a more robust infrastructure at the very least a useful thing to have.
That leaves the question of whether Australia should build for near or far horizons. We'll leave that one to economists, if only because they're even more likely to be zealots than technologists! ®