Queueing models based performance evaluation approach for Video On Demand back office system

Abstract

Since queueing is a common behavior in computer systems, analysis of queueing models become a mature theory that is widely applied in performance evaluation. In our work, a queueing model based approach is used to manage the performance attributes of a Video On Demand (VOD) back office system, which has been deployed and put into service on a province-scale cable TV network. Extracted from Software Performance Engineering (SPE) methodology, the approach is more compact and efficient due to its two phases evaluation processes, which are design evaluation used to estimate performance measures for the analysis of design improvement and deployment evaluation used to to determine the hardware requirements for the establishment of high QoS. As a significant research and practice, the approach can be used in any development processes of complex software systems, which have same features of collaboration, distribution and third-party integration as our VOD back office system.