Autonomic internet-scale workflows

The complexity of business processes based on SOA and Web Services is continuously increasing. Autonomic Computing may reduce this complexity by autonomously managing the systems and the running processes, their states and evolution. This paper discusses the design aspects of an autonomic Workflow Management System (WfMS) able to self-manage processes based on service composition, reducing human intervention to the minimum, in the phases of process definition and execution. Autonomic Workflows are able to modify themselves to react to opportunities or anomalies generated at run-time. To this end, they are based on semantic languages for service descriptions and user-defined policies for driving the reactive behaviour. Autonomic actions are performed triggering ECA rules, stored in a knowledge base, whose actions impact on a WfMS management interface that allows for dynamic changes of the running process