Design And Implication Of Mini-files Storage System Based On Hbase

Nowadays, our society has come into an age of cloud computing. With the speed of data production becoming even faster than before, a great deal of them have been accumulated and collected by companies of all walks of life. As a result, it comes to a big trouble to store and deal with such a massive data set. Hadoop, as an effective tool on storing and managing massive data, has become one of the hottest topics in the cloud computing technologies. HBase is a subproject of Hadoop, which generally use the HDFS as its basic file system and is designed to offering online services based on massive data. As a typical distrubite NoSQL database, HBase has many excellent features on performance, data safety, hardwear-tolerance and scalability. And now, it is widely used by companies all over the word.With the analysis of the present situation, the thesis summarizes today’s hot topics of cloud computing technology, defines the meaning of cloud computing, concludes its system structure, and makes it clear that the reasons, driving forces and the direction of the technology reform. Then, based on the study of Hadoop system structure and NoSQL theory, we get the detailed introduction of the most popular system structure on distribute system which is used to manage massive data. And, the study also tells the theoretical preparation as well as the technical foundation of HBase, which determins the HBase’s functions and its inner structure. Besides, according to the pre-introduction, the thesis proposes the ideas about how to optimize the HBase system, and gives several simple and effective optimization methods.The thesis designs and implicates an experiment of massive mini-files storage system based on hbase. We make out a detailed solution, use the secondary index to design the rowkey of HBase’s table, and then, complete the whole project by java programing.