Local Monitoring Granularity in Distributed Systems Nodes

摘要

Monitoring process is the core of the overall performance management in any distributed system node as it stays in the beginning of the control cycle - from monitoring to planning and to enacting procedures. It determines the effectiveness of planning and control decisions and system actions as well as the performance cost measured as by the generated system overloads. This paper presents a systematic scheme of the approaches to local monitoring process of a system node. The information precision of the current system state is orthogonal to monitoring system overload. We describe the linking category between these two concepts as monitoring granularity and present a systematization of possible approaches which are based on the concept of granularity. Further development of this taxonomy will be the system-range monitoring based on homogenous or heterogeneous local monitoring and case studies of applicability and effectiveness.