Service aggregation through rate adaptation using a single storage format

Abstract

Service aggregation schemes like batching, caching and rate adaptation can enable large scale interactive multimedia services, such as movies-on-demand and popular television programming. These applications are characterized by a large user population and a high density of access. Rate adaptation is a dynamic service aggregation technique that can merge streams and reclaim resources at run-time. Existing schemes for rate adaptation are expensive since they require multiple content storage formats or special hardware, therefore, this technique has not been exploited so far. We propose a novel data placement and disk scheduling strategy which supports two different content progression rates originating from a single storage format, and that is applicable to single-disk architectures as well as to multi-disk architectures with large or small stripe-sets. The proposal leads to a feasible solution for dynamic service aggregation via rate adaptation, thereby making large scale multimedia services more economical.