The rise of video streaming is dramatically affecting the Internet, according to a two-year study of Internet traffic trends that Arbor Networks recently presented to the North American Network Operators Group. Two years ago, Internet traffic was distributed evenly among a dozen Tier-1 network providers, but today the majority of traffic flows through direct peering agreements among large content providers, content delivery networks and ISPs. Consequently, Tier-1 networks have shifted their business models from simple packet delivery to richer cloud computing and content hosting services, and new players Google and Comcast have joined the top 10 list of Internet traffic producers — and the more traffic they put on the Internet, the more control it gives them over your online experience.
Traffic is growing much faster than the 50 percent year-to-year rate found by studies such as the Minnesota Internet Traffic Study; yet the “exaflood” of video traffic hasn’t drowned the Internet because network operators have found more efficient paths. The dramatic shift in traffic patterns has to do with the rise of what Arbor calls “the Hyper Giants,” 30 large companies that contribute 30 percent of Internet traffic. Thanks to YouTube, Google alone is responsible for 7 percent of all the traffic on today’s Internet, which puts it in the privileged position of prioritizing its VoIP and video calling services over YouTube without FCC permission.
The onslaught of video is also changing the nature of peering agreements. Traditionally, peering and so-called transit were very distinct from a revenue perspective: Peering agreements were “settlement free” arrangements in which packets changed hands between networks of roughly equal size and scope, but money didn’t. Fee-based network interconnects were confined to “transit agreements” in which a large network operator connected a small player to the entire Internet for a fee; peering is also strictly a “one network to one other network” arrangement. The new wrinkle is “paid peering” agreements in which a large operator permits direct connection for a small fee. Paid peering replaces transit fees that run $2-9 per Mbps with direct connection at $1-3, and enhances service, according to an article on Bill Norton’s “Ask Dr. Peering” web site which explains the value of Comcast’s paid peering and its potential collision with net neutrality regulations:
Paid peering provides better performance than transit, since the traffic takes a less circuitous route. Paid peering allows Google competitors to more easily compete with Google on performance and price without having to reach Google scale.
But paid peering may be forbidden by Question 106 of the FCC’s proposed Open Internet rules because it’s essentially two-tiered network access, Norton points out.
Paid peering illustrates how hard it is to write an anti-discrimination rule for the Internet that doesn’t have harmful side effects for all but the largest content networks. Paid peering is a better level of access to an ISP’s customers for a fee, but the fee is less than the price of generic access to the ISP via a transit network. The practice of paid peering also reduces the load on the Internet core, so what’s not to like? Paid peering agreements should be offered for sale on a non-discriminatory basis, but they certainly shouldn’t be banned.
Video is rising on the Internet, with more of it coming from legal sources such as content delivery networks and less from piracy-oriented systems like eDonkey and BitTorrent. Regulators need to look before they leap into wholesale bans on practices like paid peering that enable the Internet to carry increasing volumes of traffic. The FCC’s last net neutrality order (issued against Comcast in 2008) was an unintentional gift to purveyors of pirated content because it banned P2P throttling; going forward, the FCC should be at least as kind to network operators coping with the rise of video traffic by creative means.
Richard Bennett is a research fellow with the ITIF with 30 years of network architecture experience.