Cluster Management Research Based On LSF-Design And Implementation Of High Performance Computing Cluster Management System Based On Web

With the rapid development of information technology,the demanding of information processing capabilities has increased in all areas of national economy, particularly in the the rapid growth of high performance computing.The cluster which has the greatest impact on the current high-performance computing industry greatly improves the development of high performance computing and is used not only in the science research but also in other various fields. To ensure the system's stability, the rich hardware and software resources of cluster are essential for the system to be real-time monitoring, to balance the loadings, to get the fault information of hardware and software timely, and then to settle it in the task scheduling completing.The thesis mainly focuses on how to manage the high performance computing clusters based on LSF (Load Sharing Facility). Using LSF cluster management softwares, high performance computing cluster integrating a large number of computing resources could manage all the resources and services effectively. In most cluster systems, administrators need to monitor and control the cluster computing resources and services instantly, while users could get services and information conveniently and quickly.From this thesis, a web-based high performance computing cluster management system has been proposed using the LSF cluster management software. It could be used to manage resources,to compute cluster operations,and to achieve on-line management and publishing of high-performance computing applications.This thesis mainly contains the following aspects:1. The research in high performance computing and high performance computing clusters. The popularity of high-performance computing accelerates the development of high performance computing clusters.Understand the demands of software and hardware equipments advantages and outstandings in highperformance computing, then make full use of cluster resources for computing tasks.2. The research on LSF software.Study the working principle and structure characteristics and the cluster resources and operations management methods of LSF,then understand the concepts of the cluster management deeply.3. The research and analysis on high performance computing applications. Understand high performance computing applications in the operations of high performance computing clusters,then publish high performance applications software as a service to end users.4. The techniques research and implementation of web high-performance computing cluster management. A web-based cluster management system which is easy to be used by administrators and users has been accomplished using a variety of web technologies.5. The test and improvement of web-based high performance computing cluster management system. According to automation stress testing, the inadequacies of the Web system can be found.And then propose the improvement solution,and increase new functions to meet the demand.