Over the past five months, I had the privilege of co-editing – with Time Warner Cable’s Jason Weil – a report on Internet interconnection by the Broadband Internet Technology Advisory Group (BITAG) that we released this past weekend. BITAG’s technical working group (a multistakeholder collection of technologists) authored the report, aiming to provide an accessible introduction into how networks connect themselves together on the Internet to make “a network of networks.”
Alaskan Senator Ted Stevens caught some flak within the last decade for calling the Internet a series of tubes. While that may not be the most accurate portrayal of what the Internet is from a technical perspective, it’s not far from the truth at a higher, abstract level. It’s especially apt for Internet freedom advocates like CDT, which work to insure that traffic in the tubes of the Internet is treated equally, just as water working its way through a city’s plumbing is not routed selectively at the atomic level. Tubes necessarily need plumbers and plumbing to keep things running, but what does the plumbing of the Internet look like? How does it work? While I’ll try to give you some flavor for this here, our report serves as a crash course in Internet networking (for poets, so to speak!).
Tubes necessarily need plumbers and plumbing to keep things running, but what does the plumbing of the Internet look like? How does it work?
As mentioned previously, the Internet is a network of smaller networks. These individual networks – like your home network – connect with other networks to form a much larger system where data from one can be sent to and received from others, potentially across great physical distances. “Interconnection” is the term used to refer to how these individual networks speak and interact with one another. It encompasses both physical interconnection – how a network’s cables and hardware connect to another’s – and logical interconnection – once a physical connection has been established, how two networks route data between themselves and other networks.
Interconnection takes two forms: peering and transit. Transit is where a network offers Internet connectivity as a service to other networks, or users (like me and my home network). By buying transit, I can buy the ability to send and receive data to the entire Internet. Peering is a bit different; when two networks agree to “peer” with one another, they are agreeing to send data between their own networks, not to the entire Internet. With peering, a primary motivation is that network can avoid paying transit providers for traffic that they send over the peering link between their network and another network. For example, if I wanted to send an instant message from my home network, on Verizon, to a friend on another network (let’s say AT&T), normally that message would have to go out of Verizon’s network, through their transit provider and then down to AT&T, where both Verizon and AT&T had to pay to send and receive that data. If Verizon and AT&T peer, they can both avoid those transit costs as they don’t have to use their transit provider to send or receive that data. (In the BITAG report, we discuss in detail other motivations that on balance result in networks preferring to peer when possible.)
The Internet of today is very different from the Internet of the 1990s; there are vastly more users, high volumes of traffic, and a very different competitive landscape where there are few options for high-speed broadband Internet access and many options in the middle of the network for getting data to where it needs to go. The Internet is much more highly interconnected today; in the past, networks connected in a hierarchical fashion to larger networks and then on to one (or a few) “backbone” providers that provided long-distance transit. Today, there are many more methods for interconnecting and considerable pressure to reduce costs by reducing transit costs through peering (or content delivery networks, a.k.a. CDNs, which we discuss extensively in the report).
Interconnection has been a hot topic recently with very public disputes between content providers like Netflix and Internet service providers like Verizon, AT&T, and Comcast. Content providers like Netflix argue that they shouldn’t have to pay to deliver content to users when users have paid their Internet bill and rightly expect to get what they paid for (high-speed broadband). ISPs have argued that content providers, principally Netflix, choose to deliver traffic over links that require balanced flows of traffic, which is unrealistic for high-definition video where the ratio of data received from Netflix to that sent to Netflix can be 80 to 1. The difficulty in getting to the bottom of these kinds of disputes motivated BITAG to produce an educational, descriptive report that provides an accessible technically-grounded basis for discussing interconnection.
The FCC has regulated discrimination in local ISP networks under its Open Internet rules, but it has also recently considered whether or not it should extend these protections a bit farther up into the network, to points of interconnection. As that process unfolds, I hope that a level-setting educational report like this one is useful for all stakeholders to serve the goal of collectively crafting the set of rules that ensure the Internet is as open, innovative, and free as possible. That’s what we’ll be doing.