Cashing in on caching: on-demand contract design with linear pricing

Abstract

There has been increasing interest in designing and developing highly scalable infrastructures to support the efficient distribution of content. This has led to the recent development of content-oriented network architectures that rely on on-demand caching. This paper addresses the question of how a cache provider can monetize its service. Standard cache management policies such as least recently used (LRU) treat different content in a strongly coupled manner that makes it difficult for a cache provider to design individualized contracts. We propose the use of timer-based caching for the purpose of designing contracts, which allow providers to monetize caching. We focus on on-demand request-based contracts that allow content providers (CPs) to negotiate contracts at the time that requests are made. We propose and analyze three variations, one where a contract is negotiated only at the time of a miss, and two where contracts are negotiated at the times of both misses and hits. The latter two differ from one another according to whether pricing is based on cache occupancy (time content spends in the cache) or on request rate. We conclude that the first one is least preferable and that the last one provides the provider greater opportunity for profit and greater flexibility to CPs.