Interconnection and Competition Among Asymmetric Networks in the Internet Backbone Market

Abstract

We examine the interrelation between interconnection and competition in the internet backbone market. Networks asymmetric in size choose among different interconnection regimes and compete for end-users. We show that a direct interconnection regime, Peering, softens competition compared to indirect interconnection since asymmetries become less influential when networks peer. If interconnection fees are paid, the smaller network pays the larger one.Sufficiently symmetric networks enter a Peering agreement while others use an intermediary network for exchanging traffic. This is in line with considerations of a non-US policy maker. In contrast, US policy makers prefer Peerings among relatively asymmetric networks.

Competitive resource sharing by Internet Service Providers

“…Various aspects of paid ISP peering in presence of competition for customers have been analyzed by [18,33,35], and [12]. Our work differs from this literature in three significant ways.…”

“…The work on ISP peering can be divided into two broad categories: one that focuses on the interaction of various modes of Bill-and-Keep peering with transit (e.g., [2,12,28,44]), and another that contrasts Bill-and-Keep peering with paid peering. Our work belongs to the second category, and we focus on related work for the same.…”

“…Our work differs from this literature in three significant ways. First, these papers (e.g., [12,18,33,35]) mostly focus on market structure in presence for competition for customers and do not worry about strategic routing of the traf f ic once peering is decided upon. That is, these papers take the routing of traffic as exogenous.…”

“…Second, most of this work (e.g., [18,33,35]) takes access pricing as exogenous in order to focus on competition for customers. The focus of our work is endogenous determination of access prices, similar to [12] where providers bargain over settlement. Third, much of this literature (e.g., [18,33]) assumes that the providers are symmetric in their cost structures.…”

“…This equal and constant marginal cost assumption results in fixed and symmetric access charges for the other provider's traffic. We analyze a more general case where the providers are not symmetric, similar to [12]. However, unlike [12], who do not have any notion of traffic dependent cost, we consider general trafficdependent cost functions.…”