Quality of Service Aspects and Metrics in Grid Computing

Grid computing promises to become the future computing paradigm for enterprise application after having shown to be a quite e#ective computing paradigm for resource-intensive scientific applications. Large scale grids are complex systems, composed of a large number of components belonging to disjoint domains. Planning the capacity to guarantee quality of service (QoS) in these environments is a chal lenge because global Service Level Agreements (SLA) depend on local SLAs, i.e., SLAs established with components that make up the grid. These components are generally autonomous and join the grid as part of a loose federation. This paper investigates some of the relevant issues that must be considered in designing grid applications that deliver appropriate QoS: definition of metrics, relationship between resource allocation and SLAs, and QoS-related mechanisms