Providing Persistency Guarantees to Services Coordination

This paper introduces an approach for providing persistency guarantees to services coordination. Persistency ensures fault tolerance by storing the execution state of a services coordination using logging techniques. In the presence of faults, recovery strategies return the services coordination to a consistent state built by accessing a persistent execution history.Our approach assumes that given a services coordination expressed as a set of execution units and an order, it is possible to associate to it personalized fault tolerance properties in an orthogonal way. These properties are defined by persistency contracts associated to the execution units. The main contribution of this paper is to address persistency in a fine manner according to services characteristics and to application requirements.