A Chemical Approach for Autonomous Service Computing

Worldwide interconnected computing resources offer a tremendous amount of computing power largely under-exploited. Programming this emerging global computing platform raises new challenges that need to be addressed. Resources are heterogeneous, geographically distributed, and offer a very disparate level of reliability and appearance/disappearance rate. To make such a platform a usable computing infrastructure, it is crucial to study new paradigms aiming at making it autonomous and self-reconfigurable to automatically adapt itself to changing conditions.

Chemical programming has been recently identified as a promising paradigm to design autonomic systems. Within such a paradigm, computations can be seen as chemical reactions controlled by a set of chemical rules. In other words, data are molecules of a chemical solutions, reacting together to produce new data. Reactions take place in an implicitly parallel, and autonomic fashion. HOCL is a chemical language implementing these concepts as well as providing the high order: a rule is seen as a molecule and rules can be replaced like any other molecule. The chemical programming paradigm relies on a multiset to store the molecules.

Our objective is twofold. First, we work on providing a new chemical language adapted to the service composition problem and expressing its execution on a distributed platform through this paradigm, taking advantages of its implicitly parallel and autonomous properties. Second, we are working on a distributed chemical engine bringing such concepts to reality and giving birth to the next generation of systems for service computing.