Abstract
The Internet service provider market is very competitive. Small and medium-size Internet service providers (ISPs) are competing for customers, while, at the same time, they are under price pressure from upstream providers. Therefore, these ISPs have to reduce their overall cost of interconnection. In order to address this issue, Internet exchange points (IXPs) have been built up in recent years, which allow small and medium-size ISPs to go into public or private peering with other ISPs. However, those ISPs do not have sufficient information to select the optimal set of ISPs, with which they should go into private peering agreements. In this paper, we describe an approach, which provides ISPs with the information about the most economical interconnections to other ISPs. This approach helps small and medium-size Internet service providers to reduce their interconnection costs for upstream connectivity and to improve network performance for their customers. To achieve that, our approach uses Internet topology information in close neighborhood of the ISP (which is determined by the set of ISPs connected to the IXP), measurement information about the number of bytes transmitted, and traffic pricing schemes. Based on real data, our analysis results demonstrate that our approach provides the necessary information to ISPs for locally optimizing their interconnection agreements (e.g. peering, sibling, transit agreements).