Resource Management for Cloud Computing.

Cloud Computing will surpass the Internet in importance; it is a relatively new technology model of Internet-based computing, whereby information, software, storage, servers and networking are provided on demand. According to different end user requirements, such as data or computing intensive applications are applied in cloud environment. Cloud computing makes computing as a utility and has the potential to transform a large part of the IT industry, making software even more attractive as a service and shaping the way in which hardware is designed and purchased. We review this cloud computing technologies, and indicate the main challenges for their development in future, among which resource management problem is pointed out and attracts our attention. Combing the current resource management and broker theories, we present new strategy of how to manage resources to deal with different user requirements of cloud services. We described the related work of resource management and broker architecture in this thesis, accomplishing four major research issues. Firstly, we present cloud resource management framework and service-oriented broker to efficiently utilize potential physical resource such as network, storage and computing resources. This framework provides on-demand service to users and optimizes resource utilization by employing virtualization technology. It has the ability to enhance the system performance and satisfy the many of user request. The service-oriented broker is presented in this framework. It is aim to discover, select and reserve the network and end-system resources. It is able to provide guaranteed service by reservation mechanism to meet user’s demand. The several scheduling algorithms are employed for finding best match of combined resources. Secondly, we address the problem caused by failure of job submission in cloud environment. We present dynamic resource selection algorithm under user’s QoS requirement which would benefit for selecting the best combined resource for real-time multimedia applications. This algorithm is aim to reduce the job fail rate and efficiently utilize network resource to enhance system performance. Thirdly, we solve the virtual network resource allocation problem in cloud computing. We present virtual network-aware resource scheduling algorithm for real-time multimedia applications that is in order to efficiently utilize physical network resources. This algorithm will reduce the physical network traffic and guarantee the service request of the users. Furthermore, the streaming data will be transmitted from selected data source, and then application performance will be maintained. Fourth, we address meta-job scheduling problem for computation application in cloud environment. We present V-heuristic scheduling algorithms for allocation of virtualized network and computing resource. The objective of these scheduling algorithms is aim to achieve high system throughput, improve load balance and minimize meta-job processing time. From the practical aspect, we develop a simulator which is extended from CloudSim simulator. This simulator offers a simulated environment to model this resource management framework. The GloudSim only concentrate on end-system resource without getting concerned about virtualization technology and network resource, so that they can accelerate study progress of new cloud technologies