Methods and tools of distributed computing systems metamonitoring

This paper discusses the problems of monitoring a distributed computing environment that includes a lot of loosely-coupled heterogeneous computing resources. The principal distinctive feature and novelty aspect of the developed metamonitoring system first of all consists in creation of the universal software agents, capable to collect data about state of different environment compo-nents from local monitoring systems, to unify and analyze these data, and if necessary to generate and execute automatic control actions directed on changing equipment working mode, including the shutdown in case of emergencies. The approach is based on web-technologies and multi-agent technologies usage, methods of expert systems creation, methods of decentralized processing and the distributed storage of data.