Towards Distributed Monitoring and Performance Analysis Services in the K-WfGrid Project

The complexity and the dynamics of the Grid environment and of the emering workflow-based applications on the Grid require novel performance monitoring and analysis services in order to capture monitoring data at multiple levels of abstraction, to analyze the data and to correlate metrics among entities. In this paper, we present the design of distributed monitoring and performance analysis services in the K-WfGrid project. We give an overview of the architecture of the performance and monitoring services and discuss useful performance and dependability metrics for workflows in K-WfGrid. We describe basic system components including the monitoring and instrumentation service, the performance analysis service along with data representation and service interface.