The Design And Implementation Of Visualization System Based On MapReduce Application Under Grid Environment

As the grid computing plays a more and more important role in HighPerformance Computing area, the applications in grid computing grow faster andfaster. After Google develops and prompts the MapReduce computing architecture,deploying MapReduce application under Grid environment will extend the applicablefield of grid computing. As the technology develops, more and more users get toknow and want to use grid products. However, the bottleneck is the highperformance products which based on Command Line Interface are known as it’scomplex to operate. To solve the problem, the demand of developing a visualizationsystem based on MapReduce application under grid environment has been raised.To show the complex computing architecture to user, the thesis chooses a own-developed commercial visualization framework named Common Widget. It includespowerful component library and has excellent user experience after comparingseveral current visualization frameworks. An important problem in visualizationsystem is the bad performance. In order to cut down the time when data transfer frombackend to web, the thesis uses JNI, Ajax and related technologies to improve thewhole system’s performance. For common user, manually deploying the MapReduceapplication and visualization system into environment is not easy. After investigatingthe range of users who use grid applications, the thesis choose Linux x64platform asthe deploying environment. Then use RPM packaging technology to directly deploythe whole MapReduce solution into user’s environment.The thesis divides MapReduce application workload into three types:Application, Job and Task. And then show the list and functions of these workloadsonto visualization system hierarchically. The thesis not only implements all theMapReduce application’s functions, but also creates several easy-to-use andpersonalized functions. Moreover, to make different types of user to focus onspecified functions, the thesis configures different levels of right limit. The welldesigned MapReduce application’s visualization system not only makes thecomplicated product easy to handle and reduces the company’s cost, but also lowersthe operational user’s threshold of professional skills, so that popularizes the product.