A Time-Window Based Request Distribution Strategy for Media Server Cluster

Characteristics of streaming media service is different from the traditional networked services, e.g. web service, which makes designing the request distribution strategy face additional challenges. By exploiting the high skewness in streaming media access pattern, a variable-duration time-window based scheduling strategy is proposed for clustered streaming media systems, and it can help to improve cache utility while keeping workload balance among the servers in the cluster. The throughput requirement of storage subsystem is relieved, and the system utility is improved. Finally, computer simulations are performed to investigate its performance, and the results show that our proposed algorithm is better than traditional scheduling strategies.