Parallel and Distributed Stream Processing: Systems Classification and Specific Issues

Roland Kotto-Kombi (1) , Nicolas Lumineau (1) , Philippe Lamarre (1) , Yves Caniou (2, 3)

1 BD - Base de Données
2 LIP - Laboratoire de l'Informatique du Parallélisme
3 AVALON - Algorithms and Software Architectures for Distributed and HPC Platforms

Deploying an infrastructure to execute queries on distributed data streams sources requires to identify a scalable and robust solution able to provide results which can be qualified. Last decade, different Data Stream Management Systems have been designed by exploiting new paradigm and technologies to improve performances of solutions facing specific features of data streams and their growing number. However, some tradeoffs are often achieved between performance of the processing, resources consumption and quality of results. This survey 5 suggests an overview of existing solutions among distributed and parallel systems classified according to criteria able to allow readers to efficiently identify relevant existing Distributed Stream Management Systems according to their needs ans resources.

Mots clés: Distributed Stream Management Systems; Map Reduce; Workflow

Domaines: Base de données [cs.DB]; Calcul parallèle, distribué et partagé [cs.DC]

Roland Kotto-Kombi, Nicolas Lumineau, Philippe Lamarre, Yves Caniou. Parallel and Distributed Stream Processing: Systems Classification and Specific Issues. 2015. ⟨hal-01215287⟩