Research And Design Of Hybrid Cloud Storage Architecture

As an application model of cloud computing, cloud storage provides data storage solutions with strong availability, good scalability and flexibility, based on server clusters, grid computing, distributed file system technology. However, once data is outsourced to the public cloud storage providers, users lose the control of it. The concerns users have about public cloud storage are data reliability and availability, vendor lock-in, data security, and network transmission performance, etc. Meanwhile, the private cloud storage running on data centers of enterprises, will cost high on hardware and infrastructure. The deployment of it is difficult to be maked once and for all. It needs continuing maintenance because data growth and unexpected problems are inevitable. All this mentioned above become the obstacles to the further development of cloud storage.The hybrid cloud storage architecture studied in this paper is a combination of the private cloud and public cloud storage. It can provide users with better data security and access performance, taking full advantages of private cloud. It can also meet the sudden outbreak of storage needs in enterprise business, offering necessary storage resources without the need for additional infrastructure. Thus, the research and design of hybrid cloud storage architecture, which should provide a seamless and transparent data access on the clouds, having reasonable and efficient data storage and migration strategy, with more secure and available data storage service is significant in the development of cloud storage.The main work in this paper is as follows:Firstly, the storage solution based on information dispersed algorithm in hybrid cloud environments and its data security and data availability are studied. After that a hybrid cloud storage architecture is presented. In the architecture, the strategy of data storage, retrieve, migration and access statistics is designed. The service provider selection problem is defined when spread out the data among a set of public cloud storage service providers, then an algorithm to solve the problem is proposed, designed to achieve good tradeoff among storage cost, algorithm cost, vendor lock-in, transmission performance and data availability.Secondly, on the basis of the hybrid cloud storage architecture proposed in this paper, a small business-oriented hybrid cloud storage system is designed and implemented. This paper will show the deployment architecture of the system and the overall design philosophy including functional division, class level and database. Java technology and a variety of development frameworks are used to complete the system’s implementation, the process of which will be introduced briefly in this paper on different functional modules including user management, data control, cloud service provider management and system management.Finally, the hybrid cloud storage system is deployed in the test environment with Hadoop’s HDFS as an implementation of private cloud storage, and Amazon S3, Windows Azure, Aliyun, Qiniu as the providers of public cloud storage. System functional testing and performance testing are taken. Experiments results demonstrate that the hybrid cloud storage architecture proposed in this paper can complete stable data access operation while the data storage and retrieve strategy are effective in reasonable amount of time in the prototype system. Also, the provider selection algorithm is efficient and accurate to find optimal data placement solutions.