The Design And Implementation Of HPC High Performance Monitoring System

With the rapid development of the social informationization, the high-performancecomputing technology has become a mature technique, computer cluster technologysuddenly appears on the horizon, and in this way we can also meet the requirement forthe high-performance computing. For now cluster technology has increasingly occupiedan high proportion in this field. The IBM Platform HPC high-performance clustermonitoring system is a complete solution for high-performance cluster monitoring andmanagement. Platform HPC reduces the costs for user to manage cluster, and suppliesreal-time remote monitoring and cluster management for the user.This paper explains concept of the high-performance computing, clustermonitoring system and some related technology, then analyses the realistic demand formonitoring services in high-performance cluster applications, gives the architecture andfunctional design of host list monitor, performance, blink/unblink led, VNC these fourcore modules, then gives the data analysis and achieves this high-performance clustermonitoring system. And take test for all the core modules and in the end of this papergives the test result report of HPC high-performance cluster monitoring system.The test results of system indicates that with this monitoring system thehigh-performance cluster system can get a better performance, reliab ility and scalability.The high-performance cluster monitoring system meets the requirements of users andget the predetermined design goals.