Many opponents of the NBN like to argue their point using a car analogy: why should the government buy everybody a Ferrari, they say, when all that horsepower will be wasted when a Hyundai Excel will get you to work just as quickly? After a few weeks actually using 100Mbps, I agree that they're right in principle, except that the NBN is less like an Excel than it is an eight-seater Kia Carnival people-mover.
First, some background. As regular readers may know, broadband speeds have been changing rapidly at the Braue house: from 2Mbps last October to 35Kbps in November, 18Mbps cable just before Christmas, and most recently to a 100Mbps cable service thanks to an offer to test out Optus' Supersonic Broadband service.
That last one initially sounded like being gifted a Formula One hot lap, but actually, living with 100Mbps has proven only that those speeds, not to mention the 1Gbps maximum so controversially announced two days before the election, are all but irrelevant in today's day and age.
This is not a surprise. I have argued repeatedly that the NBN isn't about speed, but about providing a lowest common denominator that service delivery firms can count on when building new offerings. What was a bit of a surprise, however, is that the internet world basically isn't ready for 100Mbps, even where it is already available in Tasmania or Telstra's Point Cooke. If you live anywhere else, you won't get those speeds until your friendly neighbourhood NBN Co person rolls past your door sometime in the next decade.
You can read all about my benchmarking here, but suffice it to say that you'll only get 100Mbps if you're accessing content hosted in the same city, on the same ISP's network. Servers in Sydney cut speeds to 78Mbps, and pulling data from a server in Brisbane brought my 100Mbps connection down to 22Mbps. If I wanted to access a server in Perth, I was getting data at just 9Mbps.
Worse still, accessing overseas websites (as I'm sure we all do every once in a while) was throttled right back to the 2 to 3Mbps zone because each in-transit router hop adds additional latency, and due to the inviolable strictures of our trans-Pacific umbilical cords. Even using BitTorrent, which is among the most bandwidth-hungry applications you're likely to find these days, I was unable to leech content faster than around 2MB/s (approximately 18 to 20Mbps, which represents the aggregation of multiple individual data streams).
The NBN is less like an Excel than it is a Kia Carnival eight-seater people-mover.
You can throw all the last-mile bandwidth you want at this sort of service, and things won't get any better. These speeds may be a dream for those of you struggling with ADSL at 2 to 3Mbps, but it wasn't substantially different than the 18Mbps cable service I already had.
The thing is, I didn't really mind. Most of the things I do online now happen so quickly that I don't really need them to come faster. Videos stream quickly; software updates download in the background and only tell me when they're ready to be installed; the odd video-conference is smooth and, generally, reliable.
So, does this mean the NBN is a waste of time? Absolutely not; most people aren't getting anywhere near 18Mbps, and giving many Australians a faster connection will be like installing plumbing in a house where slop buckets and bedpans had previously been the only options. The NBN isn't about me, or you; it's about all of us.
The key here is to think of 100Mbps as an aggregate figure: factor in your FetchTV or TiVo set-top box downloading on-demand content in the background; you watching streaming videos in one room; your two children watching a streaming movie upstairs; and your significant other doing an online three-way in the other room (an after-hours work video-conference, I mean); and it quickly adds up.
This is the kind of usage that we can anticipate five years from now, and it's why the NBN is more like a Carnival than a Ferrari or an Excel. The NBN makes possible, things for which ADSL will simply fall over. And while 100Mbps services may not be worth subscribing to right now, you'll value the headroom down the track.
The somewhat unimpressive reality of 100Mbps services represents an inflection point for carriers ... just providing faster pipes isn't enough.
The somewhat unimpressive reality of 100Mbps services represents an inflection point for carriers. If they want to help customers get the most out of the services they're selling, just providing faster pipes isn't enough. They're going to need to match these services with high-quality content delivery networks hosted on distributed networks with local servers in all the cities where their customers are.
Think about what Akamai has been doing for some time, what AAPT and others are doing, and what innovators like FetchTV are relying upon: providing a hosted content service that's run completely on-network, so things like latency and congestion are minimised. As they wait for ubiquitous NBN services to be rolled out, ISPs should be pushing forward with content partnerships and focusing on the one thing they can control: optimising their networks' routing tables so they can minimise latency.
The wide variability in overseas services and network topographies means I can access servers in sub-Saharan Africa faster than servers in New York City. This defies intuition but makes technical sense: the more routers your data traverses, the slower it moves. NBN Co knows this, and surely must be factoring it into a network build that is about much more than speed. Labor may be selling you the car based on the width of its tyres, but in the end the real measure of the NBN's success will be how many passengers it can carry and how well it can manage congestion on tomorrow's information superhighways.