The Research And Implementation Of Web-Based Cluster Monitoring And Management

Currently , The cluster system has become a development tendency for the high performance computing architecture , which greatly improves the development of high performance computing ,and gradually developed from scientific research area to other fields. In practical application , the software and hardware resource of cluster are vfery richs,which need to be monitored in real-time to help cluster system achieve load balancing and complete the scheduling task successfully , and solve the failure of software and hardware timely to ensure the stability of the system .Firstly , this paper introduces the computer cluster and related technologies , taking LSF (Load Sharing Facility) as an example , it elaborated the importance of load balancing software in the cluster management and the existing problems , and further illustrates the importance of real-time monitoring software for the load balancing of cluster management ; Secondly , this paper analyzed the current cluster monitoring and management technology,and pointed out the problems in the current cluster monitoring and management system;Again , presented in detail the design and implementation process of the web-based computer cluster remote monitoring management system (RTM),by using the LSF and Cacti based on network traffic monitoring graphics analysis tools . And then , according to the system's functional components , we analized some of the main module of the system , such as user login management , management module,cluster system monitoring management module , computational operation monitoring management module , cluster resources graphics drawing module , cluster resources alarm tip module , and so on . Finally , we tested and doing data analysis for the Web-based cluster monitoring and management system, summarizeed the deficiencies and seeked the improvement methods for the system , and puts forward the direction of future research .