Internet Architecture Evolution and the Complex Economies of Content Peering

  • Authors:
  • Bill Krogfoss;Marcus Weldon;Lev Sofman

  • Affiliations:
  • Alcatel-Lucent's corporate Chief Technology Office (CTO), Plano, Texas;Corporate CTO for Alcatel-Lucent and also a member of Bell Laboratories;Alcatel-Lucent and a member of Bell Laboratories, Plano, Texas

  • Venue:
  • Bell Labs Technical Journal
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

In recent years, the Internet has been evolving from a hierarchical (multi-level) network to a flatter structure, due in large part to the exponential increase in end user video consumption. This evolution of the Internet has resulted in more highly interconnected ISPs and content providers, with tens, hundreds, and even thousands of inter-domain connections. Each domain or autonomous system (AS) connects to many others through so-called peering connections aimed at reducing expenses and improving performance in a highly competitive on-demand video world. The economics of settlement-free “peering” with neighboring Internet service providers (ISPs), compared to buying “transit” from a backbone provider were attractive several years ago. However, we will show that this approach has created “dis-economies” and loopholes in the Internet ecosystem leading to unbalanced economics. This paper will look at the operation and architecture of the Internet today and how economics have played a strong role in the evolution of the topology. Peering was intended as a reciprocal relationship between ISPs used for symmetric traffic exchange without fees being passed between peering partners. However, as transit fees grew due to increasing video traffic, asymmetric peering with content partners (known as “content peering”) emerged as an alternative peering arrangement. Content peering relationships grew dramatically over the last several years and, as a result, transit expenses fell substantially from several hundred per month dollars to under $10 per Mb/s per month today. While this peering provided temporary economic relief for ISPs, the longer-term consequences were not anticipated. This paper makes the argument that large content providers have used their dominant traffic position (e.g., Google may represent 10 to 20 percent of transit traffic) to mandate content peering. ISPs faced with significantly increasing costs due to large volumes of content providers' traffic were compelled to accept asymmetric peering, in which more content was sourced into their networks than was sourced by their networks. This asymmetric peering was positioned by content providers as mutually beneficial to the ISPs but our analysis shows that the long term impact of content peering is negative for the ISPs. We have developed a representative economic model for the Internet which is used to provide insight into the cost and revenue distributions for different types of ISPs. This model is used to clearly and quantitatively demonstrate the economic imbalance that exists between ISPs and large content providers. © 2012 Alcatel-Lucent. © 2012 Wiley Periodicals, Inc.