Font Size: a A A

The Research Of Cloud Storage Technology Based On OPENSTACK

Posted on:2015-01-25Degree:MasterType:Thesis
Country:ChinaCandidate:Z H XiongFull Text:PDF
GTID:2268330428990998Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
With the advent of the era of big data and mobile Internet, theinformation is showing explosive growth, in order to effectively manageand utilize vast amounts of data on the Internet, cloud computingplatform-based cloud storage technology in recent years has been rapiddevelopment. By clustering applications, grid technology, or distributedfile system implementation of cloud storage technology has become aneffective tool to deal with massive data. Currently more and more peoplefocus on cloud storage,those research which related on cloud storage andcloud computing has become a hot spot and future trends.OpenStack is one of the most popular open source cloud computingplatform that can effectively simplify the deployment of cloud platforms,different users convenient and efficient to deploy their own private cloud.OpenStack can provide cloud infrastructure services like Amazon EC2andS3cloud storage, and its scalability is very good, through customizeddifferent sizes, different access requirements clouds, to meet thedifferent dimensions, different requirements of the computing or storageneeds. Now many world-renowned manufacturers have already providedOpenStack-based cloud computing or cloud storage service,manycorporations in the domestic such as Sina,Iqiyi,360,etc.have builddomestic large-scale commercial OpenStack cloud deployments.Now thecloud able to provide services outside has more than twenty.The latest version of OpenStack has seven major components, namely,Compute, Object Storage, Dashboard, Identity, Network, Image Service andBlock Storage components. This paper studies the cloud storage service, so we focus on the OpenStack Object Storage and Block Storage components.In the cloud storage clusters, the problem which must be addressedis the load balancing of cluster nodes. In this paper, I combine thecurrent popular OPENSTACK cloud platform and study the architecture andimplementation of SWIFT stored objects deeply, as the objects stored inall the services provided by the external node must be achieved througha proxy, which may lead to the problem of uneven access to each node andthen affect the performance and stability of the storage system. Theexisting versions of OPENSTACK have not given the proxy node loadbalancing solution for this problem. According to the problem, we combineWEB server load balancing scheme and proposed an improved weighted leastconnection algorithm, which can effectively load on each agent nodedynamically adjusted so that the load of the whole system is relativelybalanced. In the study of cinder block storage, the metadata from cephmanagement strategy based on directory replication and directorymigration dynamic load balancing strategy departure, I add heterogeneousnodes, I/O performance and network performance considerations, proposedthe load balancing strategy based on the visit latency and the realizethe effective management of Cinder metadata.
Keywords/Search Tags:Openstack, cloud storage, load balance, metadata management
PDF Full Text Request
Related items