Tuple-based Technologies for Coordination

By tuple-based technologies we refer to any coordination system that uses associative access to shared dataspaces for communication / synchronization purposes. The idea of using a shared dataspace to coordinate concurrent activities first appeared in the Linda coordination language, which defined a coordination model based on the so-called tuple space, as well as a set of primitives, that extend a host computational language, to access it. The basic Linda model has been extended in various ways by different languages / architectures in order to deal with the different requirements of different application areas, from high speed parallel computations (the context in which Linda was designed) to Internet-based multi-agent architectures. This chapter surveys various tuple-based coordination systems and models, introducing a taxonomy (based on the enhancements with respect to the original Linda model) as well as a set of criteria to classify the considered projects. Our aim is not just to supply an updated reference to existing tuple-based coordination systems, but also to provide the reader with some helpful guidelines to compare coordination models and systems. Given the book focus, we will restrict our survey to the technologies that are explicitly targeted to open distributed systems, trying, at the same time, to offer a perspective that is as wide as possible.