Constraints-Aware Scheduling for Transactional Services Composition

Abstract

Composite Web services need transactional support to guarantee their consistent and reliable execution. Due to the long running and inter-organizational characteristics of Web services, current approaches for transactional Web services composition adopt compensation mechanism to maintain atomicity. A common assumption is that a compensation operation can be applied at any time with no cost. However, compensation operations are typically associated with temporal and cost constraints, which make compensation mechanism problematic in this new environment. To address this problem, we distinguish two types of scheduling for transactional Web services composition: time aware scheduling and cost aware scheduling. We devise several algorithms for scheduling, which can ensure the atomicity of composite services when compensation operations have temporal constraints, and assist composite services to maintain atomicity with minimum compensation cost when compensation operations have cost constraints. We benchmark our algorithms by simulations and the results show that our algorithm decreases the compensation cost and in turn improves the QoS of transactional services composition.