Design And Implement Of The Heterogeneous Resource Monitoring System For BESⅢ Distributed Computing

BESⅢ experiment is a large scale high energy experiment.It produces PB massive data every year.The local cluster has exposed its weakness to process such a large amount of data.Therefore,a distributed computing environment has been built which combined cluster,grid and cloud resources.The computing environment integrated the heterogeneous resources from the cooperative organizations all over the world and provided computing and storage service together for the BESⅢ experiment.The resource in the BESⅢ distributed environment has a high degree of distribution and heterogeneity.To ensure the stability of the resource,a uniform monitoring and management system is needed.The system should collect information from various sources and manage the resource automatically.Such a monitoring system has been designed and implemented in this paper.The main work is as follow:(1)Investigate the resource status about the resource monitoring for the distributed computing of high energy physics and learn some advantages from other monitoring system.Considering the feature of the BESⅢ distributed computing environment,the uniform system architecture is designed for BESⅢ resource monitoring and management.(2)As to the information collection,with passive monitoring,all the available information sources outside the system can be fully used to collect monitoring information,so that can ensure the comprehensiveness of monitoring.Meanwhile,active monitoring is developed to test the availability of the resource and collect availability status information.(3)As to the resource management,a policy-based management method is presented following the access control language XACML.The policies contain the interpretation of the monitoring information.The policies are used to analyze the collected information and evaluate the resource status.Then the control actions can be executed according to the results of the policies to manage the resource automatically.(4)As to information access,a user-friendly web portal is provided to expose the monitoring information.The information is shown in a layered way.Users can get the overview of the resource at first glance through summary information and obtain the interested detail information.Together with the real-time information,historical information is also provided.