A Framework for QoS-based Resource Brokering in Grid Computing

Effective and efficient exploitation of Grid computing facilities requires advanced resource management systems to automatically and transparently ensure the fulfillment not only of functional requirements but also of non-functional ones. This paper presents a framework for brokering of Grid resources, virtualized through Web Services, which can be dynamically configured with respect to multiple syntactic and semantic description languages and related matching strategies. Hence, it discovers and selects resources and automatically allocates application tasks to them on the basis of both functional and quality of service (QoS) requirements. In particular, the paper presents a framework specialization which aims to select a pool of resources whose overall performance allows for satisfying time and cost constraints for the execution of an application partitioned in concurrent tasks according to the data parallelism pattern.