Dynamic Checkpointing of Composite Web Services

Abstract

Web services provide services to their consumers in accordance with terms and conditions laid down in a document called as Service Level Agreement (SLA). Web services have to abide by these terms and conditions failing which, SLA faults result. Fault handling of web services is a key mechanism using which SLA faults can be avoided. We propose fault handling of choreographed web services using checkpointing and recovery. We propose checkpointing in three stages: design, deployment and dynamic checkpointing. We have presented first two stages of checkpointing in our earlier publications. In this paper we discuss the need for dynamic checkpointing and, various factors to be considered while revising checkpoint locations dynamically. We also propose a framework for implementing dynamic checkpointing.