On Distributing the Runtime of the Chemical Programming Model
Marko Obrovac, Cédric Tedeschi · HAL (Le Centre pour la Communication Scientifique Directe) · 2011
Internet is nowadays a global computing platform comprising myriads of autonomous services which require targeted composition and coordination. Nature-inspired, and more specifically chemical programming models, in which a computation is seen as a set of reactions between molecules interacting freely in a solution, has emerged as a promising alternative for programming such platforms. While much works recently highlighted the versatility and expressiveness of such a model, its distributed execution is still a widely open problem. With this paper, we start the study of a distributed execution environment for chemical programs. We propose a framework based on a peer-to-peer network on top of which molecules and reactions are distributed. We exhibit some optimality properties of our algorithms. A real-world prototype has been developed, and deployed over the nation-wide Grid'5000 testbed. These experiments confirm the viability of our proposal.