The Deployment Of Mass Storage Systems

Mass storage system is a popular storage solution which has solved the problems of large amount and concurrent access to the data. Mass storage system is constituted of a set of computer nodes that forms a cluster. The cluster has two different types of nodes, control nodes and data nodes.A cluster must be deployment before use. Cluster deployment is not a new subject, it is something that people have studied for a long time. Because of this, making it more difficult to find something new in it. However, the old solutions cannot keep up with current technology. Cluster, the ancient concept, now showing new kind of sight. The old technology cannot solve new problems. Nowadays, there exits many kinds of clusters, the existing cluster deployment solutions cannot solve the problems in the deployment of new coming clusters.On this basis, this paper presents a source code-based deployment solution that can fundamentally solve the problems in cluster deployment. And starting from this point, we design the mass storage cluster deployment tools called MSSDT.MSSDT consist of the scripts, configuration files and source code packages. The configuration files are responsible for describing and storing the characteristics of entire cluster and the composition and of nodes in the cluster. The scripts describes the package building process and the installation process. The source code is the core of MSSDT.The configuration file is divided into generic configuration file, shared by all nodes and node configuration file that describes the characteristics of a single node. Scripts are used to produce the cross-compiler tools, installation package, system package, deploy the server and install the systems on the nodes. Source code packages include the basic functional packages, development tool kits, and special purpose packages.Finally, due to lack of time and various other reasons, some ideas has not been implemented yet and some are not completely implemented. In short, the ideas of deployment presented in this paper is very worthy of further study, and based on that we can implement more comprehensive deployment tools.