The Desigh And Implementation Of High Performance Cloud Storage System

With the development of information technology,the amount of data in today has been exploding.Traditional network storage,such as NAS,has limitations in terms of system scalability and access methods due to the limitation of hierarchical data model and protocol interaction,which can not meet the needs of increasing data storage and exchange.Cloud storage can break the traditional lack of network storage,allowing users to access storage services over the network anytime and anywhere.Based on the data model and interface defined for Amazon S3 and the BOSS,a distributed storage system independently developed by our lab,this paper implements a set of efficient and reliable cloud storage system.The main work of this paper includes:1)A detailed study of Amazon S3 data model and interface,design and implementation of a high-performance Amazon S3 gateway.At the same time,it designed and implemented data access control mechanism for Amazon S3 model.2)This paper deeply studies the metadata management of cloud storage.It analyzed and studied the advantages and disadvantages of metadata management strategies such as dynamic subtree segmentation.It managed the namespace management by exploring the use of NoSQL database3)This paper provided the directory hierarchy to users on the flat object storage model.The directory traversal,directory renaming and other challenging issues in-depth analysis,put forward the corresponding solution.4)Conducted a detailed performance test of the implemented BOCS cloud storage system,and conducted an in-depth discussion and analysis of the existing problems and optimization solutions.