The Case for Microcontracts for Internet Connectivity
MetadataShow full item record
This paper introduces microcontracts, which are contracts for "slices" of the Internet connectivity along dimensions such as time, destination, volume, and application type. Microcontracts are motivated by the observation that Internet service providers carry traffic for different classes of customers that use the ISP's resources in a variety of different ways and, hence, impose different costs on the ISPs. For example, customers have little incentive to move less important traffic from a peak time interval unless their contract reflects the ISP's costs in that time interval. To address this inefficiency, microcontracts divide connectivity into fine-grained units so that prices more directly reflect the costs that the ISP bears for delivering the connectivity at that time. We explore the feasibility of applying microcontracts in realistic Internet service provider settings by characterizing the traffic patterns from a transit network along two specific dimensions: time-of-day and distance travelled. We argue that microcontracts are both feasible and advantageous to both buyers and sellers of Internet connectivity. We develop a model to help ISPs derive customer demand functions from observed traffic patterns; using this model, we show that making contracts for Internet connectivity more fine-grained can improve the aggregate gain of an ISP and its customers.