| It is rapidly developing of information and communication technology, computing modelexperiences these phases: First, send jobs to large-scale processors together; Then, develop tothe distribution task process model which is based on network; Now, it becomes the CloudComputing model of processing on demand. Cloud Computing is a new IT model, with thesupport of Cloud Computing Center which is of large scale, automation and high reliability inthe back end, we can access to all kinds of Cloud applications and information convenientlyonly needed to access to the network, and there is no need to install and maintain by ourselves.Currently, the whole world is busy studying Cloud Computing, building Cloud ComputingPlatform and related applications. Many companies provide services through CloudComputing, for example, Google’s search service is one of the Cloud services.The search service has become a necessary tool in our daily work, study andentertainment. Through the use of search engine, the retrieve ability improves a lot. So, thesearch engine is the successful combination of modern computer, network and traditionalsearching theory.In Cloud Computing, search service needs to migrate tasks to other machines to get abetter application performance and service effect in many times. And in this article, wepropose a framework and algorithm of live migration of virtual machine.Live migration of virtual machine is one of the most popular virtualization technologies,which is also a unique technology of virtualization. Live VM migration migrate a virtualmachine from a host to another host, and make sure that there is no effect to the applicationsrunning on it. Through live VM migration, search service in Cloud Computing will not beinterrupted. So users will not feel the activity of migration.This article studies related technologies of Cloud Computing, search engine, livemigration of virtual machine and ant colony algorithm in depth. Based on these, we design alive VM migration framework for search service oriented, and propose a job schedulealgorithm based on live VM migration. We decide when to migrate through the comparison oftime consumed, which target node to select. Migrate the VM on node with heavy workloadswhich performs badly to a proper node by using the technology of live VM migration canachieve load balancing. The algorithm mainly contains load balancing strategy, informationstrategy, trigger strategy, location strategy and selection strategy. With the comparison ofexperiments’ results, it is found that the migration mechanism we employ here is fit for searchservice in Cloud Computing.The first chapter introduces the background of the article, researches and analyzes the Cloud Computing platform of the world in detail.The second chapter reviews the related technologies, for example, the conceptions andarchitecture of Cloud Computing, etc. It also describes the technology, characteristics andclassification of search engine, live migration technology and several methods of livemigration, and ant colony algorithm.The third chapter proposes a framework of live VM migration, and introduces thestrategies of live VM migration.In chapter IV, we optimize the task scheduling algorithm. Adjust the performance offramework through static method, and we propose a two level queue deadline drivenalgorithm for task scheduling to improve the performance of cluster and the experiment’senvironment, process and analysis of results, and varies the correctness of the algorithm.Chapter V introduces the work of this article and points out the research directions in thefuture. |
PDF Full Text Request