Skip to Content
Uncategorized

Peer Pressures Could Strain the Web

A fight between two big Internet carriers, Comcast and Level 3, reveals how private business disputes might imperil the Internet.
December 6, 2010

It’s easy to assume that the Internet is unbreakable: that it’s a collection of overlapping networks so resilient that universal connectivity is simply a given. It’s not—as a battle between two large carriers of Internet traffic could soon remind everyone.

The feud began soon after Netflix hired a company called Level 3 to deliver its TV shows and movies over the Internet. Level 3 runs the biggest and most interconnected of the 36,135 autonomous networks that collectively make up the global Internet, but even Level 3 can’t deliver shows directly to couch potatoes who want to watch The Tudors on a Wednesday night. For that, it needs to hand off Netflix’s digital bit streams to Internet service providers such as Comcast. And that’s where the problems start.

By one estimate, Netflix now accounts for over 20 percent of U.S. Internet traffic into homes during prime time. Other network experts doubt that number, but everyone agrees that video downloads are a huge and growing source of Internet traffic. Seeing the surge in video traffic, Comcast told Level 3 in November that it deserved to be paid for delivering the shows to viewers who get their Internet access from Comcast. Although Comcast didn’t explicitly threaten to cut its subscribers off from Netflix unless it got paid, it technically could do that as a negotiating tactic.

“It’s a new world emerging,” says Andrew Odlyzko, a University of Minnesota math professor and expert on the inner workings of the Internet. “Comcast is beginning to swing its weight around.”

Level 3 argues that it shouldn’t have to pay Comcast, because the only reason it puts the traffic on Comcast’s network in the first place is that Comcast’s customers ask for it.

While this is an unusually public fight, spats like it are as old as the Internet. Normally, the private networks that make up the Internet strike confidential deals to trade traffic. Those deals determine the routes that information travels as it moves through the Internet. If AT&T and Sprint, for example, strike a peering deal, then traffic can flow between their two networks directly and quickly. Otherwise, AT&T might hand a Web page off to Verizon, which would then deliver it to Sprint. Money often changes hands, too, with smaller and weaker networks paying, and the bigger networks generally collecting money to grease the flow of data.

The history of fights between big networks indicates that one of two things will soon happen in the Comcast-Level 3 fight. Either the two companies will privately settle their differences, or they will start an all-out war that balkanizes the Internet—what is known in the trade as “depeering.”

That has happened when negotiations break down and two networks refuse to trade traffic, directly or indirectly. Back in 2005, Level 3 got into a battle with Cogent Communications, another huge global network. One day, Level 3 got fed up with Cogent’s refusal to pay it the money Level 3 felt it was due, so the company simply stopped exchanging Internet traffic with Cogent. Suddenly, Level 3 customers and Cogent customers couldn’t reach each other’s Web pages—unless they had other ways of hooking in to the Internet.

The Internet was soon alight with angry customers from both companies. Many declared they wouldn’t pay for partial Internet access. Two weeks later, Level 3 essentially caved, as the two companies agreed to trade traffic at no cost to either party. (This is known in the trade as “settlement-free peering.”)

In 2008, two other ugly depeering battles flared up. The first cut off much of Scandinavia from many U.S. websites for two weeks. The second severed full Internet access to major U.S. and Canadian universities, the state government of Maine, the U.S. court system, and millions of Sprint mobile customers. That fracture lasted three days.

Big networks abide by different peering policies of their own design, but AT&T’s and Verizon’s are typical. Verizon, for instance, insists that settlement-free peers send and receive roughly the same volume of traffic: the ratio may not exceed 1.8 to 1 in either direction. It’s an odd tradition, on its face, but it’s based on the assumption that traffic should be balanced across networks over time. If a network wants to send Verizon substantially more traffic than it receives from Verizon, then Verizon makes it pay. And even if a network receives more traffic from Verizon than it sends out, then Verizon also can make it pay for allowing the imbalance. “No one has come up with a rigorous way of measuring value,” Odlyzko says.

Now the industry’s tradition of insisting on balanced traffic flows is smacking into the huge growth of Netflix and other video services from companies such as Apple and Amazon. Video downloads—huge files streaming in one direction—quickly throw traffic ratios out of whack. In the past, depeering has proven to be so painful for both sides that the companies involved quickly find a way to heal the breach. But it’s possible the flood of video surging onto the Internet will undermine the global network’s tendency toward peace.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.