Adaptive atomic capture of multiple molecules

摘要

Facing the scale, heterogeneity and dynamics of the global computing platform emerging on top of the Internet, autonomic computing has been raised recently as one of the top challenges of computer science research. Such a paradigm calls for alternative programming abstractions, able to express autonomic behaviours. In this quest, nature-inspired analogies regained a lot of interest. More specifically, the chemical programming paradigm, which envisions a program's execution as a succession of reactions between molecules representing data to produce a result, has been shown to provide some adequate abstractions for the high-level specification of autonomic systems. However, conceiving a runtime able to run such a model over large-scale platforms raises several problems, hindering this paradigm to be actually leveraged. Among them, the atomic capture of multiple molecules participating in concurrent reactions is one of the most significant. In this paper, we propose a protocol for the atomic capture of these molecules distributed and evolving over a large-scale platform. As the density of potential reactions has a significant impact on the liveness and efficiency of such a capture, the protocol proposed is made up of two sub-protocols, each of them aimed at addressing different levels of densities of potential reactions in the solution. While the decision to choose one or the other is local to each node participating in a program's execution, a global coherent behaviour is obtained. We also give an overview of the course of execution when a program contains multiple rules and provide a rule-changing mechanism. The proof of correctness, as well as intensive simulation results showing the efficiency and limited overhead of the protocol are given.