A Common Carriage Model for Peering & Transit

Net Neutrality advocates come in several forms. Some argue that the future of the Internet is at stake and new regulation must be put in place or the dominant ISP incumbents will gouge all of us with higher charges. Other advocates insist that the real issue has been the change in FCC policy going back to fall of 2005. They say that the FCC created this mess by choosing not to enforce the rules that have been traditionally imposed on the telephone oligopoly. Namely, the notion of common carriage is no longer being enforced by the FCC, or the "powers that be" – in Washington D.C. The debate over net neutrality and how to “fairly” create a multi-tier Internet has been full of sound bites and bumper sticker "wars". Both sides seem to represent false choices that will ultimately resolve themselves into either complex new anti-trust regulation or an unsustainable economic model whereby broadband providers will see increases in their costs without a commensurate improvement in revenues.

Regardless of your position on the issue, most everyone would agree that the traditional notion of common carriage enables free market economies to flourish. Discriminatory carriage practices like those imposed by the railroads at the turn of the last century created monopolies. In the U.S. government’s anti-trust action against Standard Oil, it became clear that the collusion with railroads to provide itself with secret rebates, and thus a much lower cost means of distribution, was one of the key ways in which Standard Oil came to dominate the industry. In essence, by controlling the distribution of kerosene, the railroads and Standard Oil colluded to restrict free trade. By thwarting the competition through the control of distribution of a critical commodity, the railroads enabled Standard Oil’s monopoly. In other words, collusion between distributors and suppliers hampered these industries until anti-trust action unleashed the market to once again run free. Let’s make sure we don’t see a repeat of the distribution-supplier collusion in a multi-tier Internet or we’ll see governments invoke tools like the Sherman Anti-trust act. On the other hand, if governments preemptively regulate the Internet we’ll see another problem, which has its roots in how the Internet’s essential service – bandwidth – was commercialized.

Today’s wholesale model for IP “bandwidth” comes in two choices – peering and transit. Peering is based upon the notion that two parties will receive approximately equal value by connecting their IP networks together under a settlement free basis. Transit is by definition a customer-supplier relationship. As such, the customer receives value from the transit supplier ISP to connect them to the Internet backbone. At the time of the early Internet's commercialization, these agreements were deemed to offer the most scalable and robust means of insuring an openly interconnected network of networks. Transit suppliers interconnected with one another as peers or as a transit-supplier and customer basis. This has now emerged into a market whereby “Tier 1” transit providers only peer with one another on a settlement free basis. Tier 2 and 3 transit providers buy service from Tier 1s and others. Content companies, like Yahoo, MSN, AOL, CNET, etc. all either peer or purchase transit. In order to peer, these content companies interconnect their networks with “last-mile” or access network providers – the Broadband ISPs. The net effect is that the "big media" companies on the Internet have an economic advantage because they can more easily peer than the small media.

In the view of peering coordinators, they must receive approximately equal value in agreeing to a settlement free traffic exchange. This makes the peering coordinator probably the most influential individual for determimg the access network's current and future profitability. This process has been popularly described as bringing “eyeballs” to the content and vice versa. Under this economic regime, content networks don’t have to buy expensive transit and potentially get much better connectivity, while access networks create a better user experience for their broadband subscribers. However, this has also created an economic problem for the access networks. They must constantly increase the capacity of their networks without a commensurate increase in revenues.

Even peering has a cost. With all the soft costs of peering, some now estimate that it is much less expensive to purchase transit than to peer for the same quality of connectivity. This may be especially true with many of the Tier 1 transit providers. Explicit and implicit service level agreements insure that robust connectivity exists all the way out to the interconnect or peering point with the access networks. Regardless, an end-to-end best-effort Internet does not truly exist. It only exists inside the access networks because there have not been uniform technical or settlement standards for providing service level guarantees from the peering point through the access network. ISTP and TCP-Mustang will create such a uniform approach. The settlement mechanisms solve the economic problem while the access network can turn their cost center into a revenue center. For every incremental premium byte of bandwidth they offer, they receive additional revenue -- just like the transit providers do for all of their bandwidth. Their peering policies can be used to right the other economic scaling challenge for best-effort service.

From a common carriage standpoint, there are two challenges for the access ISPs: discrimination and settlement fees for the premium streaming bandwidth. If the access networks, as some have proposed, discriminate against particular websites or generally restrict service to the smallest content provider, then there is a clear problem with adhering to the principles of common carriage. This could result in roughly the same situation that we had with Standard Oil and the railroads. Small suppliers could not compete with the large ones because they either must have formed a direct business relationship with them or will pay much more for the service. This would put them at a significant cost disadvantage.

Enabling a multi-tier Internet while adhering to the principles of common carriage represents a challenge primarily from an economic standpoint. In other words, how do we insure that even the smallest content provider can obtain access to premium bandwidth services while enabling larger high volume providers to obtain a discount for the services used? Under the InterStream approach, requiring access providers to set a “rack-rate” price when they sign their escrowed peering agreement enforces a common carriage model. As long as the settlement fees are set at reasonable rates, then the traditional requirements of common carriage have been met. Individuals and small content providers can purchase their required bandwidth at a reasonable rate and allow their streams to reach virtually all corners of the Internet.

Thus, the only real danger, from a common carriage standpoint, will be that the access network providers set a rack-rate price that is prohibitively expensive. Preferred rates negotiated directly between the access and content providers under the escrowed peering model will enable another pricing tier. The technical implementation of ISTP insures that non-discriminatory practices are enforced.

The question, therefore, becomes what role the InterStream Industry Association should play in influencing or setting rack-rate pricing. I’ll have more on that in a future InterStream blog entry.

Jeff Turner


Post new comment