Atomicity imposes an all-or-nothing requirement on the observable effects of a business process. In a business process, each task may need to be compensated or retried during failure recovery to assure process atomicity. The properties of whether a task can be compensated and whether it can be retried may be dynamically changed for business needs. Fro example, a company changes its decision on a supplier to obtain a lower price of products before placing a purchase order. When the decision change leads to unavailability of the product-return service, the task of purchasing is changed to non-compensable. However, if the task properties are not changed under proper control, the atomicity of the process may be violated. To address this problem, we present a rigorous approach to examine if...[ Read more ]

Atomicity imposes an all-or-nothing requirement on the observable effects of a business process. In a business process, each task may need to be compensated or retried during failure recovery to assure process atomicity. The properties of whether a task can be compensated and whether it can be retried may be dynamically changed for business needs. Fro example, a company changes its decision on a supplier to obtain a lower price of products before placing a purchase order. When the decision change leads to unavailability of the product-return service, the task of purchasing is changed to non-compensable. However, if the task properties are not changed under proper control, the atomicity of the process may be violated. To address this problem, we present a rigorous approach to examine if the atomicity of a business process is guaranteed under different possible bindings of its task properties. The proposed approach is useful as a proactive measure to prevent violation of atomicity due to dynamic changes of these properties. These concepts are illustrated by a customer service process of a computer company that provides assembly, delivery and installation of computer sets.