Control-driven coordination programming in shared dataspace

Abstract

This paper argues for an alternative way of designing coordination models for parallel and distributed environments based on a complete symmetry between and decoupling of producers and consumers, as well as a clear distinction between the computational and the coordination/communication work performed by each process. The novel idea is to allow both producer and consumer processes to communicate with each other in a fashion that does not dictate any one of them to have specific knowledge about the rest of the processes involved in a coordinated activity. Furthermore, the model is inherently control-driven where communicating processes observe state changes and react to the presence of events and where the main communication mechanism is limited broadcasting (as opposed to either point-to-point or unrestricted broadcasting communication). Although a direct realisation of this model in terms of a concrete coordination language does already exist, we argue that the underlying principles can be applied to other similar models. We demonstrate our point by comparing our model with an established and widely used coordination framework, namely the Linda-type Shared Dataspace model, and we show how the functionality of the former can be embedded into the latter, thus yielding an alternative Linda-based coordination framework.