Research And Implementation Of Dynamic Distributed Computing System For Small And Medium-sized Computer Cluster

Because of the rapid development of network technology and communicationtechnology, people find that through the network they can use a lot of personalcomputers or workstations to form a cluster system. This cluster system has flexiblestructure and high performance to achieve parallel computing, and it is not onlylow-cost but also easy to implement. Especially with the universal application ofInternet, people usually need to process massive data and information; in this case, itis particularly important for the research and development of distributed computingsystems. Since the1990s, with the explosive growth of the Internet, Internet enteredinto millions of households, and as an effective tool to communicate information andwork together, it had become to an essential part for people’s work and life like basicnecessities. In addition, Internet brings together tens of thousands of data resources,hardware and software resources, and the development of Internet makes sharing ofthese data become possible.This paper first analyzes the research status of distributed computing system, andthen makes a brief introduction about a variety of typical distributed computingtechnology such as Middle-Ware technology, Grid technology, and Web Servicetechnology. Based on the understanding and research about distributed computing,this paper proposes a Web Service technology framework, dynamic distributedcomputing systems based on small and medium-sized clusters, the system can solveproblems of long time-consuming when a single computer process a largecomputational problems and make large computing faster. In this paper, the design of distributed computing system mainly consists of threemain modules: the client, the server-side and the communication interface. Client andserver-side is C/S architecture, the client communicates with server-side via thecommunication interface, multiple clients can simultaneously perform a computingtask, and then return each calculated results back to the server-side. In addition tomeeting the needs of the client’s request, the server-side originally developed a set oftask management subsystem which is mainly responsible for the real-time monitoringof tasks and the client’s current state, and it is implemented based on B/S structure. Inthis paper, the implementation of the communication interface mainly uses the WebService communication.In Chapter4, this paper completes the coding of system on the basis of overallsystem design using C#.Net as a development language. The task managementsubsystem in server-side was completed by OR mapping and ExtJs, and it can make areal-time monitoring on distributed computing task completion progress and thecurrent operating status of each client through browser.In Chapter5, this paper tests the performance of distributed computing systemby way of experiment, through experiments it can be seen that when the systemprocesses large computational problems, the efficiency of the implementation hasbeen greatly enhanced, at the same time, with increase of node in cluster, theefficiency will improve a lot. The experimental results show that the dynamicdistributed computing system in this paper is feasible and it has achieved a goodresult.