Research And Application On Distributed Power Cloud Storage System

With the expansion of the state grid company’s business, the number of office filegoes up so greatly as to cause low efficiency of storage and make the management of filebecome tedious. Therefore, the research on the power cloud storage system has the veryhigh application value. The cloud storage system has widely extended and developedwhich is dependent on discussion about the concept of cloud computing in the present. Inaddition, the cloud storage system have high availability, high scalability, high reliabilityand can be integrated with various types of storage medium in order to provide access todata storage and business functions. HDFS is the implementation of the underlying filesystem of Hadoop open source cloud framework and it has the characteristic such as highfault tolerance and high transfer rate. HDFS also can be written in the form of flow orread the data in the file system, which can solve the problem of file storage and accessspeed, and realize the huge amounts of data storage and management. On the basis ofthe research and analysis about cloud storage theory, we will realize a distributed powerstorage system in order to make the system gradually become an important data storageplatform of SGC.Firstly, the thesis has analyzed and researched on the current extensively discussedcloud storage model and its key technology. Therefore, HDFS is described which is theunderlying file system of open source cloud computing framework of Hadoop.Secondly, on the basis of the research on HDFS, the thesis has carried on the deepdiscussion and research according to the load balancing mechanism and thedisadvantages of small file processing. Thus, a new load balancing optimization strategyand small file storage optimization strategy are put forward. The design of the weightedload balancing algorithm comprehensively considers the performance factor of each nodein the process of system resource allocation in order to make each node in the clusterfully play a role. Thus, load balancing can be achieved. On the basis of the traditionalsmall file processing scheme, small file storage structure and index structure optimizationhas been designed. Therefore, it greatly improves the storage and retrieval efficiency ofthe small files in HDFS.Finally, on the basis of the original HDFS architecture, this topic has integratedchapter3and chapter4optimization strategy together and realized the power cloudstorage system. The system has mainly realized the file upload, download, share, andbrowse functions. The implementation of the cloud storage system has great influence onthe deepening application in power company of HDFS.