A Resource Allocation Controller for Cloud-based Adaptive Video Streaming

Video streaming accounts today for more than half of the global Internet traffic. Content Delivery Networks (CDNs) are employed to provide scalable and reliable video streaming services. Typically, the delivery systems are provisioned to meet the expected peak demands which are due to time-of-day effects. However, such a sizing strategy may either not be able to handle unpredictable flash crowd scenarios, or lead to underutilization of the network with a consequent waste of resources and revenues. Cloud computing offers a way to match the users demand by scaling the allocated resources and by billing the service with a pay-as-you-go pricing. In this paper, we focus on the design of a control plane for cloud-based adaptive video streaming delivery networks. By employing feedback control techniques, we design a dynamical Resource Allocation Controller which throttles the number of virtual machines in a Cloud-based CDN with the goal of minimizing the distribution costs while providing the highest video quality to the user. Results indicate that our resource allocation controller is able to significantly decrease distribution costs and to provide a high video quality to the user.