Data-aware automatic derivation of choreography-conforming systems of services

We present two data-aware algorithms to automatically derive web service compositions from global specifications. We show that a natural projection oriented derivation does not work in general, since some issues arise when projecting the set of implementations from the global specification, and we show how our approach deals with them. We use the constructions of two well-known languages as basis to define our models. In particular, given a WS-CDL choreography, our algorithms automatically extract a set of WS-BPEL compliant processes such that the interaction among these processes reproduces the behavior depicted in the choreography. This is achieved by introducing some control messages which make services coordinate as expected. With respect to our previous work on this kind of derivations, the main improvement of the models and derivations given in this work is their data-awareness, that is, the introduction of variables within the model, which strongly improves the expressiveness of our previous model based on FSMs. As a result, the formal model is enriched with new constructions such as workunits, which enable the definition of a complex conditional behaviour, and the derivation algorithms are necessarily more complex and sophisticated. The new derivation algorithms are implemented in the new version of our public derivation tool DIEGO.