The Internet-based Distributed Computing System Design And Realization

Distributed computation using spare resource on the network has been an important way of high-performance computation. Providing better performance-price ratio, it is more convenient to solve complex problems than a single computer. The paper introduces an Internet-based distributed computation system to meet the demand of complex problems in biological computation.Firstly analyzing the drawbacks of some existing distributed computation systems and studying corresponding proposals, the paper then designs and implements a Java-based distributed computation system. Based on the master-worker model, the system adopts a three-layer structure and uses Java applet as the worker program. The participators can join the computation only with a Java-supported browser. The computation application will be deployed as a computation web service. Using this way, it overcomes the inconveniences of conventional computation resource usage. It has the characteristic of easy use, fault-tolerance and security.The paper deeply discusses the design and implementation of the distributed computation system, explaining the implementation of key functions, including checkpoint, fault-tolerance and task scheduling. The key technologies such as Web service and system security have been emphasized in the paper. The algorithm of motifs seeking in DNA sequence is used for the system test. The result shows that the system meets the design demand and has good practicability.In the end, the paper sums up our work during the design and development, and discusses how to improve the system in the future.