Private Cloud Storage And Cloud Computing Implementation

With the rapid development of Internet, more and more data on the Internet have been created, which introduce the hot research work on Big Data. In order to solve some problems of Big Data, Cloud Computing arises. Cloud computing, compared with the traditional way of high performance computing, has the advantage of cost-effective, high reliability and high scalability. Private cloud is a kind of cloud computing implementation approach. Compared to the public cloud it is more security and lower investment cost.Currently the research work of cloud computing is mainly concentrated in the aspect of text processing, and less in the aspect of video and image processing. Compared with text, image and video processing is more complex. But now, the network video data volume proportion in the whole network data is becoming bigger and bigger. The analysis of large amounts of video will become popular in cloud computing research area. In this paper, based on the research of private cloud platform, video copy detection is put forward and implemented based on Hadoop platform.The main contributions are listed as follows.1. A private cloud platform based on Hadoop is set up;2. A video storage system based on HDFS and a video feature database based on HBase are proposed and implented;3. Based on MapReduce framework, a video feature extraction algorithm and video copy detection algorithm are designed. Experimental results show that the video copy detection runs faster on private cloud than on the single computer; In case the size of processed data is greater than the data processing capacity of cloud per unit time, the total time of data processing has a negative exponent power function relationship with the number of compute nodes in cloud.