Monitoring WS-CDL-based choreographies of Web Services

Self-healing software, and especially self-healing web services, is one of the challenging tasks for a next future. This paper is a step in this direction, insofar as it describes a framework for monitoring and diagnosing web services. One of the main difficulties is that faults may propagate from one service to another, which makes the diagnosis a crucial issue in order to react properly. We propose to use a chronicle recognition approach, currently in use for monitoring complex industrial systems. The challenge is to extend the existing approach to the specific context of web services, the main point being to deal with highly distributed systems. We focus in this paper on choreographed web services, whose choreography model is described in a WS-CDL file. We use a simplified example of an e-foodshop web service to illustrate our proposal. A monitoring platform, named Matrac, has been developed and experimented in the context of e-commerce. This work has been achieved in the framework of the WS-Diamond European project.