Experimenting Scheduling Policies with Continuation-based Services

The ultimate success of computational grids as a production-oriented commercial platform for solving problems is critically dependent on the support of economy-based mechanisms to resource management. In market-based service-oriented grids the possibility to manage the execution of services is advisable to control their provision and to meet Quality-of- Service (QoS) requirements of both resource providers and consumers. This paper presents experiences in developing and testing a scheduler middleware as a component of a service provider infrastructure for flexible provision of grid services based on application-level service execution control. The full and programmable control of service execution provides the scheduler with flexible mechanisms to dynamically adapt the execution of services according to both the changing conditions of the environment where they operate in, and the requirements of service users. In particular, two proposed cost-driven resource-sharing scheduling policies were implemented and tested showing different behaviors with respect to the system workload (i.e. number of service requests).