摘要：In this paper we present our approach to the management of distributed systems based on semantic description of available resources. We use ontologies for a semantic description of the monitored system and other aspects of monitoring and management (such as metrics) and introduce a feedback loop on underlying infrastructure. Such an approach allows to automate monitoring and the ease the work of administrator. We introduce concepts behind a novel automatic management system, SAMM, developed within our research. We discuss the core mechanisms used in the system - the estimation of future measurements, approach to knowledge gathering, and the process of decision making. Then we provide some details on the architecture and implementation of SAMM.
