Skip to Main Content
With the increasing scale of distributed resources, which couples a wide variety of geographically distributed computational resources, storage systems, data sources and computational kernels, service management and scheduling is a complex undertaking. Since most services on the Internet are mutually independent, users have trouble in accessing several services to achieve a single purpose. Furthermore, the resources are owned by different individuals or organizations with their own policies, have different access and cost models, and have dynamically varying loads and availability. To address the confronted issues, we propose an adaptive approach in this paper for distributed services composition and allocation, which is service-centric and workflow-oriented. We use XML to describe the services by encoding such factors as resource configuration, performance, and service-specific capabilities, which enables services to be easily developed, composed, deployed, relocated, monitored and managed in distributed environments. We use service composition technology to construct composite service by combining several independent services, which can be considered as the element services of the composite service, executed as workflow process. The allocation of such element services will be decided dynamically due to the economic objective function to address the issues of adaptation, availability, performance and economic cost.
Machine Learning and Cybernetics, 2004. Proceedings of 2004 International Conference on (Volume:1 )
Date of Conference: 26-29 Aug. 2004