This paper presents a monitoring and forecasting system for Grid resources, based on Java and agent technologies. Performance monitoring, measurement and forecasting of Grid resources is an active research area, several researches and projects are studying and developping Grid information systems in order to provide users or schedulers with services necessary to determine which grid resource fulfills specific task requirements. Computational Grids are intrinsically heterogeneous and dynamic systems, these features make extremely difficult to build an information system that update itself automatically following the Grid configuration evolution. The technologies involved in our approach allow the implementation of a multi platform information system that is able to collect several node and network status data without explicit registration of new nodes or network configuration. The history of collected data is used by a resource broker to forecast the node and network performances. A graphical interface has been also implemented to show real time performance evolution and forecasting error and to allow users to select the best prediction algorithm.
Java Services for Distributed Resources Performance Monitoring and Forecasting / Guerriero, Andrea; Pasquale, C; Ragni, F.. - (2010), pp. 88-92. (Intervento presentato al convegno 8th IEEE International Conference on Virtual Environments, Human-Computer Interfaces, and Measurement Systems tenutosi a Taranto, Italy nel September 6-8 , 2010) [10.1109/VECIMS.2010.5609357].
Java Services for Distributed Resources Performance Monitoring and Forecasting
GUERRIERO, Andrea;
2010-01-01
Abstract
This paper presents a monitoring and forecasting system for Grid resources, based on Java and agent technologies. Performance monitoring, measurement and forecasting of Grid resources is an active research area, several researches and projects are studying and developping Grid information systems in order to provide users or schedulers with services necessary to determine which grid resource fulfills specific task requirements. Computational Grids are intrinsically heterogeneous and dynamic systems, these features make extremely difficult to build an information system that update itself automatically following the Grid configuration evolution. The technologies involved in our approach allow the implementation of a multi platform information system that is able to collect several node and network status data without explicit registration of new nodes or network configuration. The history of collected data is used by a resource broker to forecast the node and network performances. A graphical interface has been also implemented to show real time performance evolution and forecasting error and to allow users to select the best prediction algorithm.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.