Internet Service Providers (ISPs) often peer with each other at Internet eXchange Points (IXPs), using a shared switch fabric, to save on transit costs and bring content closer to their consumers. Stability and efficiency of ISP connectivity and peering relationships that form at these IXPs are crucial to long-term sustainability and scalability of the Internet. In current practice, however, the price paid for traffic exchange at an IXP is typically determined by coarse, static measures that can result in end-to-end performance inefficiency in the traffic flows. This project's overarching goal was to address this limitation by exploring economic models for IXPs and architectural innovations to make the ISP peering process faster and more automated. 

The scientific merit of the project centered around (1) developing traffic exchange and peering settlement models at an IXP (for both non-profit and for-profit IXPs) through economically efficient pricing of IXP services coupled with flexible and value-driven ISP peering relationships, and (2) exploring ways to quantify the value of a peering relationship for ISPs to enable faster and unambiguous negotiation of ISP peering contracts. This project analyzed traffic rate and/or port capacity-based pricing solutions at an IXP and showed under what conditions the resulting traffic exchange solution at equilibrium is economically efficient. Since ISPs form peering relationships in self-interest, this project modeled the peering problem from a game-theoretic perspective and characterized the efficiency of the pricing solutions in terms of the best- and worst-case performance at equilibrium, compared with the most efficient traffic exchange solution achievable. To facilitate faster and clearer peering contract negotiation, the project also developed metrics based on which two ISPs can decide whether and where to peer with each other. These metrics focused on the satisfaction of the ISP pair and the longevity of the peering arrangement, with a particular attention to the stability and performance of these peering relationships. The project introduced meta-peering tools to help and automate the ISP peering process – starting with identifying a list of ISPs that are likely to peer, writing router rules to establish traffic exchange sessions with them, and extending the service to monitor all these sessions for notifying any major outages or peering agreement violations.

The innovations emerging from the project may significantly improve the health of the ISP market with an expanded role for IXPs. The Internet is crucial to modern societies, and a healthy, sustainable, open and free ISP market is key to the long-term success of the Internet as an infrastructure. Many critical communication systems that we use in our daily lives have migrated to the Internet. Public safety, traffic control/management, law enforcement, emergency management are examples of such critical systems that use the Internet on a regular basis for operation. The project's work on the Internet's economic health, ISP peering health, impacts these societal systems by improving the speed and stability of peering relationships among ISPs. Peering among ISPs make the Internet's end-to-end paths shorter. The education and outreach activities of the project included training graduate and undergraduate student researchers, outreach to the industry, and integration of the learnings from this project into the curriculum.

Last Modified: 09/06/2024
Modified by: Murat Yuksel