Research And Implementation Of A Monitoring Infrastructure In Grid Environment

Grid Monitoring System (GMS) is a fundamental component of grid system. It is possible to build a generic GMS using SOA rather than a system-specific monitoring system. However, the existing GMSes have the problems, for example don't support adding or updating monitoring content dynamically; lack of uniform mechanism for handling and aggregation of monitoring results; lack of mechanism to support system fault alarm and automatic fault handling, leading to constrains in scalability, functionality, and performance of them. Furthermore, they are the bottleneck to improve the availability, and efficiency of GMS, as well as the robustness of grid system.ChinaGrid Super Vision (CGSV), an infrastructure used to construct the GMS is proposed in this article. According to the actual requirement of users, the article presents the novel architecture based on Grid Monitoring Architecture (GMA) proposed by Global Grid Forum (GGF), adding two sorts of services, Proxy service and Archiver service. CGSV supports dynamically deploying, updating monitoring services and provides uniform mechanisms to handle the monitoring results, and standard interfaces to access and aggregate the results. Additionally, Data publisher services inside CGSV support two data transfer modes push and pull, to serve different users.This article is organized as follows. Firstly, it introduces concepts and technologies about grid as well as the information of grid specifications and organizations. It discriminates the grid computing from the traditional high performance computing and distributed computing such as P2P systems. Secondly, the article introduces ChinaGrid support platform, CGSP, and illustrates the modules of CGSP in some detail. Lastly, the article proposes a novel infrastructure for monitoring system, CGSV, defines relevant concepts, and depicts how they are implemented specifically.