Research On Cloud Service Deployment And Management Mechanisms In Cloud Computing

As an emerging computing paradigm, cloud computing enables to provide elastic and scalable cloud services to users on demand, thus to improve the utilization of resources and to reduce the cost of cloud services. The provision of cloud services mainly depends on the fact that the cloud services can be deployed and managed effectively, which means that how to effectively allocate the virtual resources to ensure the deployment and the reliable operation of cloud services.Cloud services deployment in the cloud computing environment has the characteristics of multi-user and massive heterogeneous infrastructure resources. These characteristics brought many challenges to cloud services deployment. According to cloud services deployment at different stages, the challenges involve in virtual machine images acquisition, resource allocation as well as the resource adjustment. In order to deploy the cloud services and ensure its reliable operation, this paper researches the key technologies of cloud services deployment at different stages. The main content includes the following four aspects:(1) VM images efficient acquisition method based on multi-point collaboration. With the on-demand resource provisioning of IaaS, to fetch a GigaByte sized VM image from a repository over the Internet incurs long transfer time, thereby adversely affecting request service time. To address this problem, we propose a VM images acquisition method based on multi-point collaboration. We first take advantage of high degree of commonality of VM images to divide them into fine-grain chunks by designing a mixed chunk scheme. And then, based on chunk-level sharing among different images, with minimizing the latency in fetching VM images, a VM image acquisition method based on dynamic multi-point collaboration was proposed. Finally, the experimental results show that this scheme can largely reduce request service time.(2) Virtual image transferring method based on application-friendly. With the reservation resource provisioning of IaaS, to transfer a GigaByte sized VM image cases a large of additional network traffic load which could incur the problem of data center performance sharp decline. To address this problem, this paper presents a transmission method based on application-friendly. Firstly, we model the dynamic data center network which changes over time with a kind of Time-Expanded Graph. Then, with minimizing the impact of network load caused by transferring a VM image, we formulate a programming model of VM image transferring based on the time-delay tolerance. Meanwhile, this paper employs an improved genetic algorithm to solve this problem. The experimental results show that the proposed method can greatly improve the proportionality of network links, and reduce the impact on applications performance.(3) The virtual network resource allocation method based on load-balancing. The resource bottlenecks of the hidden hops on optimal paths influence the performance of the entire substrate network and the request success rate. To address this problem, we proposed a resource allocation method of virtual network based on load-balancing. Taking into account the resource demanded by the hidden hops, this paper aims at the simultaneous loading balance of the substrate node and the substrate link, mathematically formulates the virtual networking mapping problem constrained by hops. And then, a multi-objective load-balancing particle swarm optimization algorithm was proposed to solve this problem. The experimental results show the proposed method can effectively eliminate the resource bottleneck, and provide a more balanced substrate network for the request of the consequential virtual network request, thus improving the constructing success rate of virtual network, the availability of network resources and also the profits of the infrastructure providers.(4) Resource adjustment method with coordinating maintenance scheduling and communication-cost. The physical machines are needed to conduct regular preventive maintenance to avoid hardware failures. During the maintenance time, the accommodated VMs on such servers may be migrated to other available resource. To avoid undesirable effects on underlying infrastructure and unnecessary re-migration caused by maintenance, this paper proposed a resource adjustment method with coordinating maintenance scheduling and communication-cost. To address this problem, we formulate the joint server maintenance schedule and the communication-cost VM migration as an optimization problem, and then employ the improved simulated annealing algorithm to solve this problem. The experimental results have shown that the proposed method can effectively reduce the number of virtual machine migration, the traffic caused by VM immigration and also the communication cost among VMs of a multi-tier application.