Research On Cloud Video Streaming Platform And Relative Resource Scheduling Mechanism

Cloud Computing is a new pattern of service which charges its users for what they use. It assembles magnitude resources such as computing, storage and network to form a resource pool, such that users can get services in the way as they need. Meanwhile, as a popular application, multimedia streaming is a mutual, widely accepted technique which is necessarily and required to integrate with cloud computing. However, there are still challenges when come to how to construct a flexible, stable, safe and energy saving platform, for the reason that multimedia streaming applications are widely scale, highly interaction and time sensitive.A significant character of cloud computing is virtual technique which can make use of resources sufficiently, and provide isolated, safe runtime environment. With virtual technique, cloud computing can make reasonable allocation of resources to meet the needs of clients. So it is worth to do research on mechanisms which schedule resources appropriately and make the platform load balance and energy saving.This work aims to solve problems about resource assignments and load balance of platform for VOD service based on cloud computing techniques. Focusing on these problems, this thesis carries out researches as follows:1. Survey the research background, including the concept, features of cloud computing and video streaming; requirements and feasibility of combination of these two techniques. Analyze features and technical requirements of platform for VOD services based on cloud computing techniques.2. Investigate the organization of cloud platform for VOD streaming service. A basic organization of cloud platform for VOD service is proposed, with detailed modules about resource scheduling and task assignment. Then problems to be solved about this platform are given.3. Taking advantage of VOD streaming and virtual technique, put forwards algorithms which predict the needs of VOD streaming service for memory and CPU resources of cloud platform for VOD streaming service and schedule these resources reasonably. Then test and verify these algorithms by carefully designed experiments.4. Considering the restraints of bandwidth for VOD streaming service and the difficulty of reallocation of network resources in virtual technique, design algorithm which predicts the needs of VOD streaming service for network resource of cloud platform for VOD streaming service and schedule the reassignment of network resource reasonably, according to the characters of VOD streaming service. And then test and verify these algorithms by carefully designed experiments. Finally, it proposes the scheduling mechanism for cloud platform of VOD streaming services, which includes the scheduling algorithms for CPU, memory and network resources of cloud platform as an inner method used by local scheduler and the management of building, migrating and closing proxy servers as an intra tool used by clusters.