Fault tolerance in Hadoop MapReduce implementation

Matías Cogorno (1) , Javier Rey (1) , Sergio Nesmachnow (1)

1 UDELAR - UDELAR, Facultad de Ingenieria [Montevideo]

This document reports the advances on exploring and understanding the fault tolerance mechanisms in Hadoop MapReduce. A description of the current fault tolerance features existing in Hadoop is provided, along with a review of related works on the topic. Finally, the document describes some relevant proposals about fault tolerance worth considering to implement in Hadoop within the PERMARE project in order to provide support for pervasive computing environments.

Mots clés: Hadop; mapreduce; PER-MARE; pervasive

Domaines: Calcul parallèle, distribué et partagé [cs.DC]

Matías Cogorno, Javier Rey, Sergio Nesmachnow. Fault tolerance in Hadoop MapReduce implementation. 2013. ⟨hal-00863176⟩