The use of grid computing to easily and efficiently execute data and compute-intensive applications strongly depends on new software development approaches able to separate application-domain aspects from non-functional ones, such as task mapping and deployment. In this paper, we present an object-oriented framework that is able to transparently transform non-distributed programs into hierarchical master-slave ones, and to map and schedule them onto a grid computing system. Moreover, the framework is able to leverage services delivered by the underlying middleware platform, such as resource management and communication, to satisfy user requirements. The paper presents the framework architecture, a reflection-based implementation and its evaluation atop of a hierarchical grid middleware.
A Transparent Framework for Hierarchical Master-Slave Grid Computing
ZIMEO E.
2007-01-01
Abstract
The use of grid computing to easily and efficiently execute data and compute-intensive applications strongly depends on new software development approaches able to separate application-domain aspects from non-functional ones, such as task mapping and deployment. In this paper, we present an object-oriented framework that is able to transparently transform non-distributed programs into hierarchical master-slave ones, and to map and schedule them onto a grid computing system. Moreover, the framework is able to leverage services delivered by the underlying middleware platform, such as resource management and communication, to satisfy user requirements. The paper presents the framework architecture, a reflection-based implementation and its evaluation atop of a hierarchical grid middleware.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.