Fault-tolerant Coordination of S-Net Stream-processing Networks

Fault tolerance and robustness are system properties of increasing importance, both in the domain of embedded computing as well as in the domain of high-performance computing.
In this paper we study the applicability of fault-tolerance mech- anisms in the context of stream-processing networks, in particular based on the coordination-language S-Net. We identify three basic fault tolerance mechanisms and discuss the technical solutions for them within S-Net. The applicability of these mechanisms depends on the requirement of the concrete application domain. The con- tribution of this paper is a feasibility study of tool-supported fault tolerance mechanisms in a flexible coordination language allowing for asynchronous execution. As part of this feasibility study we dis- cuss potential extensions of the S-Net language and runtime system in order to implement the identified solutions.