Research On IaaS-Oriented Self-Adaptive Resource Management Mechnisms For Cloud Computing

Cloud computing is a service-oriented business computing mode, it is put in the system of different types of physical machines and virtual machine such as heterogeneous resource integration as a virtual resource pool, according to the need to provide users with various types of services. IaaS as a base service provided from Coud Computing, is able to provide encapsulated infrastructure as service to user, by virtualized the underlying infrastructure with combined computing resources/storage and network bandwidth. Meanwhile for researching of Cloud Computing, resource management and scheduling mechanism is the keypoint towards high performance and the stability.This paper primary researched and analysed IaaS-oriented resource management and related adaptive mechanism, for adaptively allocating and scheduling virtualized resources in High Performance Computing and Web Cluster Computing field. Firstly the porpular cluster and Cloud Computing product would be discussed, combined with virtualization technology, Marketing-Oriented resource management mechanism, Negotiating policy, Advance Reservation mechanism and Adaptive mechanism, for the purpose of adaptive resource management in effective IaaS; Moreover such results would be applied to the High Performance Computing and Web cluster related environment, which lacks of efficient resource management sulotion for the peak of explosive requirement and the redundancy of over-planed capacity; Finally according to the achievement of system identification dedicated to adaptive controll system, Minimum Component Analysis (MCA) would be discussed so as to construct algorisms for adaptive resource management. This paper main job:1) This paper proposes a virtual resource oriented cloud computing resources management mechanism. Cloud computing is able to provide users with infrastructure as a service (IaaS, InfrastructureasaService) computing capacity, enables users to use efficient, reliable, economic computing resources at the same time, no additional purchase, cost of maintenance resources, thanks in large part to rely on the use of virtual resources. So how to effectively manage virtual resources, to maximize the utilization of resources and to ensure that the user use effectiveness, has become the current research problems. In order to solve this problem, this paper proposes a virtual resource oriented cloud computing resources management mechanism, based on the division of virtual resources, reserved, and scheduling strategies, to provide users with effective IaaS service. Through the simulation experiment results show that the method can improve the distribution of virtual resources and ensure that the user of the effectiveness of resource usage.2) This paper proposes a grid Web cluster of ideas. Web after using a set as a node of the cluster system resources, can effectively solve the traditional Web server load balancing and fault recovery problem, but so far, how to make more efficient use of system resources Web cluster is not the solution. In order to solve this problem, this chapter puts forward the idea of a grid Web cluster, and for Web cluster nodes after a demand management strategy, this method can effectively improve the efficiency of Web cluster on the use of resources.3) With independent management ability, this paper puts forward a kind of high performance computing method. High performance computing (HPC) clusters using the combination of a set of computing nodes using the computing power to handle different assignments submitted by scientific computing work. Inappropriate capacity planning and application of the related management mechanism, can lead to HPC cluster to a considerable number is considered to be a key factor that influence the system throughput to HPC cluster target, suspended operations. In addition, it can lead to low efficiency and waste costs. Therefore, in order to overcome these problems, this chapter puts forward the management method of an independent ability.4) This paper puts forward a kind of with independent management of the construction of the electronic commerce cloud computing power. In e-commerce, Web cluster can be combined with a series of background ability as a powerful Web server to handle heavy workload, however, the traditional Web cluster in advance to the traditional rigid infrastructure deployment and pre-configured flaws, because inappropriate plans in advance, may result in more costs and increases in redundancy and waste will consume more energy, more than the energy waste problem in the processing of rare peak. So this article puts forward the autonomous computing method to deal with these problems. First of all, we put forward on the cloud by managing virtual machine to host the new method of Web cluster, instead of on the rigid infrastructure hosting Web cluster method, then we can book in advance scheme based on a lending policy, put forward a kind of independent configuration mechanism and low-high threshold detection algorithm, in order to make the Web cluster can according to demand with other adaptive balance between cluster and recycling, improve the utilization rate of resources and effective to reduce the cost and save energy.5) System Identification is outstanding approach in the area of Adaptive Controll System against the variety of resources and uncertainty of workload. MCA is now an elegant and promising approach to system identification. So this paper focuses on developing related recursive algorithms. Then two algorithms are presented. In the first approach, rank-one modification is used to get an accurate and quickly convergent algorithm, named a-RMCA. However, its drawback is low computational efficiency. Therefore, the second algorithm is proposed to realize faster computation, named f-RMCA. Finally the simulation results verify the effectiveness of both algorithms.