A Look at Adaptability for Service Workflows

Abstract

Service-based workflow compositions like those used in Grid and Cloud computing require failure recovery mechanisms that support diverse failure handling strategies, separation of failure handling code from application code and user-defined exception handling. Apart from dealing with failures that result from such environments, these applications should be able to handle failures which are sensitive to the task context (task-specific failures). Traditional distributed system (transaction-based) recovery methods are not sufficient. In this paper, we show a selected group of projects that aim to provide exceptional behavior for service workflows. We classify these systems based on whether recover is performed at the tasklevel, workflow-level or a combination of both.