Design And Implementation Of Cloud Synchronization System

With the advent of cloud computing, the user’s data is stored in a variety of different devices, and implementing synchronization between different devices safely and efficiently is the key of the current study. In order to achieve cloud storage synchronization, Baidu cloud disk, Kingsoft fast disk, 360 cloud disk appeared recently. In this thesis several popular cloud disks were introduced at first, based on which a new kind of cloud disk with cloud synchronization system was designed and developed which is faster and safer than Kingsoft fast disk.Main work of the thesis are studying the current cloud synchronization system and analyzing the problems existing in current cloud plates firstly. Based on the achievements, a new kind of cloud disk with cloud synchronization system was designed and developed, which consists of client cloud synchronization subsystem and server cloud synchronization subsystem. They are mainly composed of monitoring module, data calculating module, compression module, transmission module and cloud memory module.The first module is the file monitoring module, which abandoned the traditional rounds of polling and was modified based on the new event monitoring mechanism through jpathwatch library. The jpathwatch is capable of detecting changes in files including deleting file, renaming file, revising file. When the monitoring module detects changes in files, the data calculating module will calculate the synchronized file blocks which are need transmitting. According to the analysis results the module determines to transfer again or transfer different data only. In order to transmit data fast and efficiently, a compression module was designed for the present system to compress the necessary data file blocks. Java language was adopted to implement Huffman compression algorithm in the compression module. The transmission module is designed multithreading, so that the files can be quickly transmitted. After the data transmitting, the system will refresh files and store files into cloud memory module in the server. The Hadoop Distributed File System HDFS was utilized to store files. Distributed storage of small files was optimized at last.The system was architected based on the hierarchical modular thought. In chapter 5, the system’s various sub-modules were tested and analyzed, and overallperformance of the system is tested, which is compared with the current popular cloud disks. The research achievements and shortcomings were summarized in chapter 6.