Video delivery services in media cloud with abandonment: An analytical approach

Distribution of video content over the Internet has drastically transformed the consumption of media. Content providers, naturally, would like to ensure that their videos play on users' devices whenever requested, without failure or interruptions. Due to the varying nature of user needs, procurement of computing resources proves to be tricky, leading to the popularity of cloud-based approach. Media cloud is a computing paradigm dedicated for multimedia services and delivers on demand services (e.g., video) by dynamically acquiring cloud resources. Use of cloud resources helps service providers to lessen their operational cost, reduce delay and abandonment rate to deliver adaptive video stream. The abandonment rate, delay, user engagement and repeat viewership plays a vital role in service providers revenues. In this paper, we use an analytical model based on queuing theory to find the effect of queue size (buffer size) on abandonment, blocking and successful services. Further, the relationship between the number of virtual machines, waiting time (delay) and abandonment rate is also examined. We also derive a relationship between the number of user requests in the system and the virtual machines required to respond to the same.