Supporting caching and mirroring in distributed video-on-demand architectures

Abstract

Distributed Videoon-Demand (DVoD) systems are proposed as a solution to the limited streaming capacity and null scalability of centralized systems. In such full decentralized architectures with storage constraints, server content management is crucial to achieving an efficient resource utilization. We have proposed and evaluated, through an analytical model, a combined scheme with caching and mirroring to manage distributed servers storage. The results show that our approach achieves the best mean service distance, independently of underlying architecture. In order to support this scheme, we have modified the storage manager of our VoD prototype so as to guarantee that the new workload of this scheme does not affect disk throughput. © Springer-Verlag 2004.