Survey on Web Services Fault Tolerance Approaches Based on Checkpointing Mechanisms

摘要

Complex business processes are based on Web services composition (WSC). Services composition dramatically reduces the cost and risks of building new business applications. Although Web services composition has been widely researched, several issues related to dependability still need to be addressed. In this aspect, one primary concern is to provide fault handling mechanisms. Over the last decade, diverse works tackling fault tolerance for WSC have appeared; most of them are based on the checkpointing paradigm. Some of the works are oriented towards orchestration and others towards choreography. In this work, we present a study regarding the different checkpointing techniques and their applicability to WSC. This study has been done considering the different types of faults (e.g. transient, intermittent and permanent) and modes of recovery (e.g. local vs global). We introduce a novel taxonomy for fault tolerant mechanisms that groups existing works according to integration approaches, fault types and modes of recovery. We present a study of the works, illustrating their advantages and drawbacks. Finally, the paper presents a discussion and outlines several open challenges regarding fault tolerance for WSC.