Design And Implementation Of Video Fragment Cloud Storage Based On Hadoop

With the rapiddevelopment of computer technology, improved Internetbandwidth in recent years, increasing numbers network users and widely used digital products, VOD application and video sharing application has developed on a largescale like catching fire. Some famous internet video websites, including Youtube,Hulu, Youku, Tudou, are born at home and abroad under such environment. At present video services turn to multi-terminal development and video data extend into great capacity with constantly rapid growth now. Today, how to provide multi-terminal users with premium video services and mass video data storage has become the hot technology among different internet video companies, and even companies of radio and TV industry also attend this grand feast of data and services.Since Google published three theses, namely The Google File System, Bigtable and Google MapReduce. cloud computing has been on the rise in IT industry and Hadoop, one of famous cloud computing framwork, comes into being. Open source characteristics of Hadoop cloud platform can reduce the cost of cloud computing researches. Moreover, Hadoop Distributed File System, the bottom-realizing module of open source clound computing software platform Hadoop framwork, has the characteristics of high mbps and error-tolerant rates.distributed file storage scheme for efficient, stable solution to the mass data.This thesis intends to design a universal distribution system based on mass video fragment data cloud storage. At first, the thesis elaborates the concept and principle of Hadoop distributed framework and describes the work principle of Hadoop HDFS in detail, which lays the foundation for mass storage of the whole thesis. Then, the thesis analyzes current conventional methods of mass video fragment storage in the industry and considers current methods and thoughts of mass video fragment storage suited to application background on the basis of actual project situation. Finally, according to HDFS video fragment storage, I design file index and multi-thread parallel process technique to improve delaying and throughout of video file systems.In the end, these theories are verified by specific project examples and the thesis reaches the final concludion from experimental data.