Is the era of the best-effort internet over? As a larger amount of web traffic is sent between a few giant players and more and more dollars are riding on the bits traveling over networks, is the public internet the best route for government files sent to Amazon or Netflix streams?
A three-year-old startup called IIX hopes that the answer is no. The firm has built what its Founder and CEO Al Burgio hopes will be the next generation of peering, but is essentially a frictionless way to bypass the public internet. The company has just raised $10.4 million in funding from NEA to help expand its vision. But before we delve into that, let’s take a quick refresher on how the internet works.
A primer on peering and the internet
The internet is not one monolithic thing. It is a series of networks, literally pipes in the ground filled with fiber, coaxial cable or copper that terminate in a variety of end devices from at consumer modems to switches and servers inside data centers. Some companies like Google might own data centers and fiber that comprise its network, while other companies might lease space inside data centers or buy bandwidth from transit companies to build out a network.
Oftentimes there’s a mix of all of this. And what we think of as the internet is the collection of all these networks, with the public internet being the highways where all packets mix and mingle on their way to a home or data center, and private networks being the closed and controlled networks owned by a company.
Today, the public interest is best effort, which means when a person requests a web page or a file, a number of different providers may work together to send the request and then return the content. But as those files become more valuable or voluminous, companies have an incentive to directly connect their networks via peering arrangements, so they can offer some sort of security and control about where and when packets get to their destinations.
Scaling peering and interconnections
Despite the public drama over peering arrangements between large U.S. ISPs and Netflix and other transit providers, most peering is done on a pretty ad hoc manner. And while internet exchange points or companies like Equinix offer neutral meeting places where a customer can buy cross connects to directly link their networks, in many cases those are negotiated on a company-by-company basis. There are a few programs where you can connect directly to Amazons cloud via Equinix for a fee without having to involve Amazon, but it’s still primarily a hands-on effort.
That’s what IIX hopes to change. Burgio describes it at a LinkedIn for peering and interconnection, where you click to connect with a company and within minutes you’ve established a direct link to their network — creating what amounts to a private network controlled by the linked parties.
While this may seem unrealistic, as of 2012 researcher and CEO of Deepfield Networks Craig Labovitz had determined that about 70 percent of internet traffic came from 150 companies, a substantial increase from about 30 percent of traffic coming from the top providers in 2009. So as the amount of traffic consolidates in the networks of a few large companies, creating this virtualized layer to connect them isn’t as far-fetched.
How it works
What IIX has done is put switches and supporting servers running its proprietary software in a variety of data centers. These locations include colocation facilities owned by companies such as Telx; CoreSite; Equinix(s eqix); Verizon(s v); Telehouse; and others. Its customers, which include Box, Google(s goog), and Microsoft(s msft), have physically connected (via a cross-connect) in at least one of these locations.
Once a customer is physically connected in any location, it can “virtually” connect to all other customers connected to the IIX PeeringCloud platform, regardless of the others’ locations using the IIX software. Burgio says that the interconnection platform is a secure Layer 2 offering, not some encapsulation scheme that might go over the public network.
The key is that in addition to IIX gear in each of the above-named locations, IIX has also has contracts with third-party Internet Exchange Points so IIX customers are also able to “tether” across the IIX PeeringCloud to participants in those third-party IXPs. It’s a similar concept to the internet, except it’s limited to the IIX participants and they can choose how and who their traffic is exchanged with using a few clicks of the web interface.
While fascinating and certainly of growing use to companies sending high-value traffic, it also has me wondering about what this means for the future of the internet. If the companies that make up the majority of the traffic create these private deals among themselves, what does that mean for the rest of the internet? Does it make the net neutrality arguments moot, given that these companies are building their own fast lanes to share traffic? Does it mean they have less incentive to support policies and practices that grow the broader internet? Burgio didn’t have a great answer for these questions, but I am curious to hear what others think.