Adaptive load sharing and scheduling schemes for distributed continuous media delivery

The purpose of load sharing is to improve both the performance and readiness of non-interrupted continuous media delivery by redistributing workload among collaborating servers. We propose and evaluate a few adaptive load sharing policies where simple scheduling schemes are derived. The objectives are: to understand the potential load imbalance problem in distributed multimedia systems; to minimize the initial latency where system workload is constantly changing; and to maintain the stability of the overall system. We develop a two-phase list scheduling with partial reordering of incoming requests to meet the challenge. The end result suggests that migrating workload to a highly loaded continuous media server generates the best implementation alternative