Diagnosis of complex Web Services (WS-DIAMOND)
Currently, fault management in Web Services orchestrating multiple suppliers relies on a local analysis, that does not span across individual services, thus limiting the effectiveness of recovery strategies.
We aim at extending Web Service composition with support for the specification, management, monitoring and recovery of complex workflows including the execution of internal activities and the invocation of external services. To this purpose, we propose a framework for Web Service composition where a Diagnoser Web Service is added to the set of composed Web Services and acts as a supervisor of the execution of the underlying workflow. The Diagnoser must identify anomalies in the execution of the composed Web Services, and explain such anomalies in terms of faults to be repaired. The identification of the problems occurring during the execution of a complex service supports a fine grained selection of the recovery strategy to be applied. The Diagnoser Web Service relies on Model-Based Diagnosis to identify the possible causes of problems.
In addition to a centralized fault analysis framework, we are investigating the possibility to carry out the diagnostic task in a distributed way within a service. When defining a complex service, we propose to add to each service a local diagnoser which relates hypotheses about incorrect outputs of the service to a misbehavior of the service itself, or to incorrect inputs from other services. A global diagnostic service is then associated with the complex service. It coordinates the local diagnosers, exchanging messages with them and, without relying on any information about the internal structure of the sub-services, it can in turn compute diagnoses at the level of the global service.
WS-Diamond Web site
Involved PhD students: Roberto Furnari