The coming glut of 100Gbps Ethernet shows that the potential growth of the National Broadband Network is limited only by the laws of physics — and the laws of Parliament.
If you've seen The World's Fastest Indian (and perhaps even if you have not), you know all about the Bonneville Salt Flats. For a hundred years, this ancient, dried-up inland sea has become a Mecca for adrenaline junkies from around the world, who gather to push machines to ever-increasing speeds across its 412 square kilometres of flat nothingness.
Network engineers have been engaged in similar efforts for decades, working long nights and days to figure out how to push data down a wire or fibre-optic cable at ever more-dizzying speeds. They may not have the live-fast-die-hard character of your average Bonneville revhead, but these information-age Burt Munros have flirted with the laws of physics and wracked their collective brains to get data from point A to point B faster than ever.
Burt Munro's modified Indian motorcycle set a land speed record in 1967. (Burtmunro1920indian image by Daniel O'Neil, CC3.0)
Their success has pushed networking to new speeds with regularity. It was just over a decade ago that carriers were using asynchronous transfer mode (ATM) technology to link cities with blistering 622Mbps trunk networks to carry unprecedented volumes of phone calls and a little bit of internet traffic. It was expected to last years, but exponentially increasing demand for data services has kept carriers investing furiously in anything they can find to boost their networks' bandwidth.
ATM quickly became irrelevant as carriers turned back to Ethernet to deliver first 1Gbps, then 10Gbps, 40Gbps and — as evidenced by a recent news announcement from Juniper Networks — 100 gigabits of data speed. And that's 100Gbps per fibre-optic cable; bundle a dozen or five-dozen fibre-optic lines together, as is commonly done, and you're talking about some serious bandwidth.
There was a time ... when they said it couldn't be done at all ... "they" were already calling time of death for Ethernet ... citing the laws of physics.
There was a time, not too long ago, when they said it couldn't be done at all. In fact, "they" — analysts, technicians, and so on — were already calling time of death for Ethernet before gigabit Ethernet had made its debut, citing the limitations of the laws of physics. Ever-smarter ways of squeezing bits onto fibre-optic cabling, however, have hushed the naysayers and kept the networking industry topping its previous achievements with impressive regularity.
Juniper's release of a commercially available 100Gbps router officially raises the bar, boosting capacity per fibre by a factor of 10 compared with the technology most carriers are using now.
Blah, blah, blah, speeds, feeds, geek talk, blah, blah, blah. This sort of stuff may make your eyes glaze over, but in the world of telecommunications it's exciting news: without this kind of technology, there will be no way to deliver projects like the NBN. It's one thing to deliver 100Mbps to every home and business, after all, but add that all up and you need a heck of a lot of bandwidth linking it all together.
Just two years ago, Telstra's Next IP network was implemented using cutting-edge technologies that push 10Gbps of data over a single fibre. Sydney and Melbourne, for example, are linked with two bundles of four fibre-optic cables each, providing 40Gbps between our two most populous cities. That's a lot of bandwidth, but replacing the 10Gbps boxes at either end of those cables with 100Gbps-capable gear would boost the Sydney-Melbourne link to 400Gbps.
These sorts of upgrades are becoming de rigueur investments as FTTH takes hold. Juniper's announcement quotes Verizon, which is especially desperate for something like this because providing its FIOS FTTH service to hundreds of thousands of homes has taxed its core networks like never before. Expect carriers around the world to quickly follow suit as a proliferation of gear from Cisco, Nortel, Alcatel-Lucent and other core network suppliers floods the market by year's end.
For network engineers, 100Gbps Ethernet is reason to break out the champagne — and cause to start working on terabit Ethernet (actually, they're already working on it). For carriers, it's another way to keep up with the ever-greater demands of ever-more-bandwidth-hungry customers. And for customers?
Well, nothing yet. But this could change quickly: the NBN will deliver fibre-optic cables running directly to a huge number of homes. Initial speeds of 100Mbps are hardly a technical limitation, but rather a convenient, round figure about which politicians can blather on and network planners can do easier back-of-the-napkin calculations when planning the NBN's design.
Building the NBN correctly will be critical if it is to be the future-proof infrastructure it can potentially be... Done right, the NBN can bring loads of bandwidth to every corner of the country and provide decades of headroom.
With the fibre in place, these fibre-optic cables could theoretically be expanded indefinitely into the future. After all, many businesses are already using 1Gbps Ethernet connections to link their various branch offices. Supporting millions of households at gigabit speeds would be technically difficult/impossible/expensive (choose your option) but with new core network equipment enabling a tenfold boost in carriers' capacity over existing fibre, it could very well be done.
That is, it could be done as long as the NBN is built correctly. And building the NBN correctly will be critical if it is to be the future-proof infrastructure it can potentially be. Yet there is a real risk that the NBN's value proposition may be undermined by cost-cutting and compromises: Stephen Conroy this week admitted, for example, that the NBN cables may run overhead instead of underground where they rightfully belong.
Ongoing dramas with Telstra's RIM-based PSTN roll-outs, which can limit access to ADSL services, have already shown the dangers of compromising on network design.
One hopes that cost limitations and pressure to deliver the NBN quickly won't compromise its underlying technical credentials. Done right, the NBN can bring loads of bandwidth to every corner of the country and provide decades of headroom.
While the Bonneville speedsters may be limited by the forces of friction and air resistance, the coming glut of 100Gbps Ethernet shows that the potential growth of the NBN is limited only by the laws of physics — and the laws of Parliament.
