Research On Resources Optimization Of Multimedia Applications In The Cloud

Cloud media combines a cloud platform and multimedia technology, which newly supports large amounts of multimedia data and provides diversified services to meet the needs of customers. A large number of multimedia access, process, and transport services and heterogeneous resources leads to more stringent QoS requirements of multimedia.So the traditional resource allocation method of media has encounted a bottleneck. This paper studies the resource allocation schemes based on the emerging challenges for cloud media resource allocation from the following three aspects:(1) This paper presents optimal VM allocation schemes based on the SaaS architecture and the VM allocation model, aiming at the two challenges of RTT and the resource cost caused by different datacenters. Specially, there are two kinds of VM allocation schemes for single-site cloud and multi-site cloud, respectively. In each case, formulate and solve the VM allocation problem to minimize the resource cost under the RTT constraints. Moreover, the improved greedy algorithm in each case can efficiently allocate VMs for the practical usage. The simulation results demonstrate that the proposed optimal VM allocation schemes can effectively achieve the minimal resource cost for multimedia application providers.(2) This paper proposes an optimal task scheduling algorithm for the dependence and priority between tasks in the cloud multimedia applications.So task-level scheduling problem based on for cloud based multimedia applications are studied. Firstly, it introduces a directed acyclic graph to model precedence constraints and dependency among tasks in the hybrid structure. Based on the model, optimal task scheduling problem for the sequential, the parallel, and the mixed structures are discussed. Moreover, combine the task node in the critical path according to the cost of limited resources. Lastly, use a heuristic method to perform the near optimal task scheduling in a practical way. Experimental results demonstrate that the proposed scheduling schemes can optimally assign tasks to virtual machines to minimize the execution time.(3) A resource scheduling and load balancing algorithm is proposed based on kinds of factors such as task characteristics of media, heterogeneity of media services and user equipment. Firstly, according to QoS features, media task and service nodes are classified into corresponding queue.Then work out resource Similarity vector and service satisfaction for each node by Resource linear normalization and Euclidean distance. Finally, allocate the service resource according to allocation vector corresponding to the maximum service satisfaction. At the same time, adjust each node using the node utilization, thereby it shorten the search time and balances the load of each node. Experiments show that, the proposed algorithm can shorten the response time, and improve the utilization rate of resources, but also improve the overall satisfaction of users.