Skip to Main Content
The use of network-based streaming applications such as YouTube and BBC iPlayer is rapidly increasing. These applications consume data at very high rates and so quality- of-service (QoS) support is necessary to ensure that they do not stall. Since the access pattern of streaming applications is highly sequential, prefetching can be employed to improve overall performance. For network access, clustering can be used to optimize prefetching. In addition, it is necessary to satisfy demand missses promptly or else non-streaming applications will be delayed. Designing a new high-performance storage system to meet these demands require the development of new analytical models. These models will be used to explore new algorithms for network-based storage that can deliver the required QoS to support streaming environments. In this paper a detailed gate-limited bulk service queueing model based on Markov chains is explored and a numeric solution is demonstrated for simple scenarios. The analysis is then applied to a high performance network-based storage server being developed here at Middlesex University. Quantitative results are presented which compare favourably with simulation.