In search of the internet: Travelling on the tubes
During the Olympics we'll be relying on the Internet more than ever. But most of us don't really understand its physical reality.
"What's so striking to me is that this is the moment when the eyes of the entire world are on London," says Andrew Blum, referring to the 2012 Olympic and Paralympic Games. "I don't mean that literally. I'm talking about the fact that people will be turning on their televisions, radios and Internet devices to follow one of the biggest broadcasting events the world has ever seen." With the exception of some of the broadcast feeds that will be via satellite, all of the Olympic news will be coming from a telecommunications infrastructure "dominated by the Internet".
New York-based Blum is an expert on the Internet. He is currently in the UK on a lecture tour supporting the publication of his new book 'Tubes: Behind the Scenes at the Internet' and has just given a TED talk in Edinburgh on the subject. 'Tubes' is a quest to find out the physical reality of what most of us perceive as being an anonymous, nebulous information access facility that pops up on our screens. It wasn't available in any meaningful sense to the world at large a generation ago, and yet today is one of the most ubiquitous technological phenomena in our daily lives.
"I like to look at these things physically," says Blum. "What you have with the Olympics is different things happening in different campuses around London. But, because of the concentration of telecommunications networks dominated by the Internet, when you speak in physical terms, the vast majority of news about the Olympics is actually transmitting through a connection of networks based in the Docklands. And this is what is at the heart of my book."
Blum's investigations into the Internet are a real journey akin to an explorer setting off to find Atlantis or Eldorado. Rather than see it as an abstract spider's web or necklace encircling the globe, he took the slightly less poetic approach of regarding it as a series of interconnected tubes that channeled information globally. "The way I see it is different from most people – I admit that – but I see it as being comprised of three dominant parts."
First there are the Internet exchange points where the networks of the Internet physically connect to each other. According to Blum there are "about a dozen of these in the world that are far more important than the next tier by an order of magnitude".
Second, there are the data centres, a term he uses to describe the actual "repositories of data, rather than warehouses full of equipment. These are located around two poles. They either cluster around us and the Internet exchange points, or they cluster way off in the boonies where power is cheapest and the air is coolest". Third, are the "lines in between. The long-distance tubes – predominantly undersea – that connect the networks".
There is an assumption among consumers that the Internet is free, that it just happens, while access to it borders on being a human right. But as assumptions go it is almost entirely misplaced. "The thing you have to remember about the Internet is that it consists almost entirely of privately owned networks. There is no public Internet, as such." Not only that, says Blum, but each of those networks is dependent on its connection to the others. "With a couple of exceptions we don't even think of the Internet as having parts and pieces. We think that when we are connected to the Internet we are connected to all of it." But in fact, we only have that sensation because of the connections made between all of these smaller networks.
What's even more striking about that, says Blum, is the way that the networks "act in their own interest, autonomously and independently. It's very hard to think of another phenomenon like that: where you have a group that needs to be incredibly co-operative, while at the same time inherently competitive: where thousands of individual pieces combine into a whole that itself has a character".
Blum's journey in search of the Internet started at his home in Brooklyn, NYC. He woke up one day to find his Internet 'broken'. At this point the 'Internet guy' was summoned and they followed the wire from the back of Blum's computer, down the back of the couch and into his garden, where they found a squirrel running along it. "I think a squirrel has chewed up your Internet," was the considered diagnosis of the technician and the matter appeared to be solved. But for Blum, who is a technology correspondent for Wired, this wasn't enough, and it led him to the conclusion that if a squirrel could chew on "this bit of the Internet, then there must be other bits of the Internet that squirrels can chew on".
Blum immediately fixed on a mental image of what would happen were you to follow the wire. And the question formed in his mind: How far can you physically follow the Internet?' Just as Blum was starting to have the idea that was to lead to his book 'Tubes', the world was falling into economic recession. In the US the government was announcing stimulus funding: "The government announced that it was going to support the physical construction of new broadband networks." There were obvious similarities between this and President Roosevelt's Works Progress Administration (WPA) in the 1930s, where there were "shovels in the ground" constructing roads. This time it would be a different type of highway.
Blum attended the launch meeting of the new programme at the Department of Commerce in Washington attended by several hundred people, all of whom owned small pieces of the Internet. "I thought as most people do, that BT or Comcast owned the Internet. As I started talking to these people I realised that, contrary to the idea that the Internet was a place that you couldn't visit and was kept under wraps for security reasons, here was a whole group of people who wanted to talk about their pieces of the Internet."
This started Blum on the more serious goal of talking to network engineers about how the Internet works and asking the fundamental question of where it was. What Blum quickly realised was that the Internet turned out to be in not very many places at all, and was in fact in a concentrated area. "It wasn't like there was a building in New York that I'd vaguely heard of might have a small piece of the Internet. The realisation was that the building in New York actually was the Internet and only one of three in New York where the networks of the Internet physically connected. There was also one in Virginia and one in Los Angeles. One in London, and Frankfurt. One in Amsterdam, and Tokyo. And when I started to poll people on what were the most important pieces of the Internet, the same answers came back."
When Blum realised that there wasn't a great deal of the Internet that he could actually visit, he set out none the less, like a tourist visiting the monuments of the capitals of the world. From his cloak and dagger, conspiracy shrouded assumptions, there emerged a mundane truth:the Internet was like an international flight path map published in the back of an inflight magazine. "It was just a bunch of tubes."
Rethinking our routing
Developing the idea of what the Internet is physically like, Blum also uses the comparison of the eco-system. The owners of the buildings that host the connections between networks are dependent on the networks interconnecting there"That's what gives them value." Correspondingly, the networks need nodal points at which to connect, and they need that point to be relatively clear, otherwise "it would be a big jumble". The largest networks solve that, says Blum, "by going to the obvious places, and at this point in the Internet's development, those places are quite fixed".
"When we send an email or make a Skype call, we assume that because there is the possibility of it going along many different routes, then it does. But the dominant routes are quite clear. Nine times out of ten the path between two points will be the same. I can guarantee that four out of five times I start a Skype call to New York it will pass through Telehouse and not Equinix. I may not know which undersea cable it's going over, but it's not as if there are hundreds to choose from."
Clearly, despite the exciting email route-tracing scenes on technology-led TV detective shows such as 'CSI' and 'NCIS', the Internet doesn't work that way. "But we assume that it does because we're so in love with packet switching. We are in love with the idea that this is the beauty of the Internet, that it allows for spontaneous connections." But what Blum discovered is that the connections are only "spontaneous" because of the fixed links agreed by "engineers over a beer and then consummated over a physical connection, usually a fibre-optic cable, usually in one of the short-listed places. Like electricity, it can only run down certain conductive pathways. It is electricity passing down tubes".
An interesting linguistic conclusion that Blum has drawn from his travels in the Internet is that it is misnamed. "The word is misleading, because it implies things that simply aren't there, and the abbreviation puts the emphasis in the wrong place." He thinks it should be called the "internetwork", but much prefers the term "internets, because it ultimately, fundamentally and inescapably is a network of networks". What we think of as a singular improper noun is to Blum something different. He also capitalises it in his book to read Internet. "We assume," says Blum, "that the definition of being on the Internet is to have the entire network at your disposal, when that simply isn't the case.
"For the most part as users, we don't need to pay attention to these distinctions. But we will do when one movement becomes so big that it removes the checks and balances of the 'inter' part. When that happens the whole balance of power shifts and there's no longer the notion of the co-operation of many networks, but the functioning of one."
Returning to the Olympic thread, Blum states that when it comes to the TV networks, we know who they are and where they are. "But the Internet isn't like that." So with all that we now know, isn't it a bit risky to depend on an infrastructure as surprisingly fragile as the Internet for one of our largest broadcasting events? "Those that know what they're doing are not depending on what we might loosely call the public Internet if such a phenomenon exists at all. They will not be simply hoping that their bits get there, or if they are then they should be forewarned that they might not. The main broadcasters will be going up-scale and buying a direct circuit from here back to New York, say. And that will be their beam of light that transmits 10 gigabits per second of data. And that's separate from what we all use. They won't be shoving it in a pipe and hoping for the best.
"Networks carry networks," says Blum. "They're all stacked. And so if the BBC wants to be unaffected by congestion it needs to keep its data on its own part of the network for as long as possible. So the onus is not on the Internet to perform, but on the individual parts of the network to make sure that the connections are performing properly. And this idea opens up a Pandora's Box because it questions our fundamental understanding of what the internet really is."
Blum is certain that if nothing else the Olympics will help get his message across. "What the games will highlight, whether or not the Internet fails during the transmission, is that the Internet isn't a single monolithic whole, but made up of parts and pieces." *
You can read Nick Smith's review of 'Tubes' by Andrew Blum on p98
Andrew Blum's five Internet surprises
Despite having found that behind his Quixotic journey into the world of the Internet was a much more mundane truth than he could ever have imagined, there were still some big surprises in store for Blum. Here are five of the biggest:
1 "The first and the biggest is simply that the Internet isn't big. What was so surprising was how short a list there was of dominant places and how small a cast there was of people operating these places."
2 "The Internet isn't a good target for terrorists. Although the shortlist of key hubs is short, there are always secondary places that act as backups. London has a handful of key sites, although one is absolutely dominant. If you were to destroy the entire complex at Virginia that would make a difference, but it would be very difficult to achieve. You might be able to take a nick out of it, but then you'd be perceived as trying to destroy the economy. So they're not tactically or strategically good targets."
3 "The importance of the engineering social network. When we load a page it appears to do so automatically. But in fact it only does so because of a very handmade process of connecting one network to another. Networks do not connect automatically: they connect because two network engineers decide it is in their best interest to connect to each other."
4 "The difference in attitude of data centre owners to sharing their technology. The most striking being the difference between Facebook and Google. Facebook has been wide open and has published the plans of its building, stating the most important route to leading the way in efficiency is to share best practice. Google, on the other hand, has said that we, their customers, aren't smart enough to understand what goes on in their data centres. It has changed since, but when I visited Google's data centre in Oregon, the Google maps image of where it was had been obscured."
5 "The physicality of the undersea cables are a surprise in itself. It turns out that satellite communication is a technology of last resort. Not for television, but for everything else. The Internet really is just a bunch of tubes. And that's the reason I called my book that."
|To start a discussion topic about this article, please log in or register.|
"Is augmented reality the next big thing or a marketing gimmick? Is it fundamental to the future or a fashion faux pas?"
- 3 LANE ROADS [01:34 pm 20/05/13]
- Define Energy. [01:25 pm 20/05/13]
- Marketing from Engineers' perspective [09:49 am 20/05/13]
- Fukushima Daiichi Unit 3 5th Floor Highly Radioactive Debris [03:09 pm 17/05/13]
- Cluster formation on cooja simulator [01:59 pm 17/05/13]
Tune into our latest podcast