Have You Ever Wondered How the Internet Works?
Andrew Blum, author of the new book “Tubes,” spent two years exploring the physical constructions around the world that enable the Internet to exist
Billions of people around the world use the Internet daily—but very few understand how it actually works. Three years ago, journalist and writer Andrew Blum set off on a journey to learn about the physical network that enables the internet to be an inescapable presence in our lives. He traveled to monumental data centers, undersea fiber optic cables and unassuming warehouses that contain crucial exchange points for his new book Tubes: A Journey to the Center of the Internet. We spoke with Blum about the Internet’s coolest (real-world) sites, the connections that make it work and where it’s going next.
How did you first get interested in writing about this topic?
I was mostly writing about architecture, but I found myself going to see actual buildings less and less, and sitting in front of my screen more and more, and that seemed like a strange way of engaging with the physical world. But even more so, I got hung up on the fact that the world behind my screen seemed to have no physical reality of its own. My attention was always divided—partly on the world around me, and partly on the world inside my screen, but I couldn’t quite get those two places in the same place.
It was just about the time of the broadband stimulus funding in the U.S., when the Department of Commerce was giving away money to encourage broadband, in 2009. I went for the kickoff announcement of the funding, and it was an auditorium full of people who had owned pieces of the internet. And that made me realize that Verizon, AT&T and Comcast didn’t own the Internet, but there were all these different pieces of it. And as I started talking to the people there, I realized that there was a way of teasing out the different parts of it, rather than having to look at it as a singly monolithic whole.
If you were to describe the physical structure of the Internet to someone who uses it, but doesn’t have a great grasp of it, what would you say?
What I usually say is that there are three major parts. There are the Internet exchange points, where the networks of the Internet physically connect to each other—and, among these, there are about a dozen buildings in the world that are more important than all the rest. The second piece is the data centers, where data is stored, and those are arranged on two poles: they’re either close to us, and close to Internet exchange points, or they’re off in the boonies, where they can run most efficiently, like in Sweden. The third part is what lies in between, the undersea cables and long-haul fiber cables and all of those that connect all the other pieces.
Of all of the places that you visited in the course of writing the book, what were your favorites?
One was Ashburn, Virginia, where a compound of buildings owned by a company called Equinix is located. It’s surprising in two ways. For one, it is one of the most important places in the Internet in America, if not globally. It’s a place where more networks connect than anywhere else. But it’s also kind of an outlier. The other places that compete with it for this title are in places you’d expect, like New York, or London, or Amsterdam. But Ashburn is a place where the Internet’s geography kind of jumps the banks and goes off in its own direction. I love that.
Facebook’s data center, in Oregon, is also an amazing place. It’s one of the few places that has tried to monumentalize the Internet—to express in architecture that it is a meaningful and important place, rather than the traditional data centers, which tend to disappear into the background as much as possible.
As you went about researching the physical geography of the Internet, what surprised you?
The thing that surprised me most was how small the community was of people who are running the networks of the Internet, and interconnecting them. When we load a Web page, it feels automatic, but in fact it only does that because of the individual decisions of two network engineers to physically connect their networks to each other. What amazed me was how social that process was—how those connections only happened when two network engineers drank a bunch of beers and talked to each other, and made that decision. Or maybe one of them paid the other, maybe one became a customer, and then consummated that decision to connect their networks by physically doing it with a yellow fiber optic cable from one router to another. The fact that that social community is so small—maybe a few hundred people—was the single most surprising thing.
Going into the future, how do you anticipate the physical nature of the Internet to change?
I think the geography is mostly fixed, for the moment—the most important places will stay that way for the foreseeable future. Certainly, our speeds will increase, because we’re demanding it. We’re not going to rest until can we not only stream HD video reliably, but we can also do it two ways, so we have video walls. I do think that’s a technology that we want, and it requires one more jump in bandwidth. It’s surprising that right now, we have these huge TVs, but there isn’t really good video conferencing on them yet. There is at the corporate level, and that’s going to start to trickle down.
Which means, specifically—and I don’t know if this is a good thing or not—we’re going to start to see our Internet bills look more like our cell phone bills, with features, add-ons, caps and things like that. This is totally against the conventional wisdom of net neutrality, but you could, for example, end up paying an extra $3 to your Internet service provider for a Netflix package, to ensure that your Netflix bits are streamed properly. Or you could pay an extra $3 for a Skype package that makes sure that your Skype traffic is prioritized when you want it to be. That’s totally anathema to the way we think about it now, but I think that is an inevitable transition in recognizing the Internet as parts and pieces, and not just a monolithic whole.