In this thesis, the idea is to use the notions of plurality and diversity, inherent characteristics of a web services environment. In order to ensure the availability of service-oriented applications, we propose an adaptive mechanism of semantics replication, where the set of services with similar functionality is a unique and fault tolerant entity, considered as opaque to other services of the composition.
The similarity of services is the key element for the substitution of one faulty service by another. We say a service S1 similar to another S2 if there is a mapping between the two functional models and their interface. There exists a substitution relationship between two Web services if they provide the same quality of service (QoS). Otherwise, services are in competition.
Thus, many providers of services can compete in an open market where redundancy can be achieved at relatively low cost (compared to a replication server within a single organization, for example). However, in order to implement the semantic replication, the replication mechanism should automatically and dynamically respond to:
1.What is/are the service(s) to be selected in order to participate in composition and/or backup scenarios in order to ensure the availability of service-oriented applications?
2.What is the recovery technique to be applied when considering service/composition reliability?