Doctorant, Polytech’Annecy-Chambery.


Courriel :  Etienne.Mauffret– @ –

Phone : +33(0) 4 50 09 42 09

Télécopie : +33(0) 4 50 09 65 59

Office : A128

Address : LISTIC – Polytech Annecy-Chambery, BP 80439, 74944 Annecy le Vieux Cedex, France

Etienne Mauffret is a PhD student at l’université Savoie Mont Blanc since October 2017. He performed teaching at Polytech’ Annecy-Chambéry and a research activity at LISTIC.

Group:  Network and Distributed Systems and Softwares

Publications :  reference on HAL



Abstract : This thesis is part of the ANR RainbowFS project. The RainbowFS consortium consists of: (i) Pierre et Marie Curie University (Paris VI), (ii) Savoie Mont Blanc University, (iii) (LIG), (iv) Telecom SudParis (TSP), and (v) Scality (industrial partner). This thesis will be carried out within the LISTIC laboratory of Savoie Mont Blanc University located in Annecy. It will be directed by Sébastien Monnet (LISTIC / USMB) and co-directed by Marc Shapiro (LIP6 / Inria-UPMC).

The purpose of the RainbowFS project is to design and implement a data management approach that provides the right level of consistency across a cloud. Existing systems require applications to use predefined consistency models. The RainbowFS project provides tools for co-designing applications and their consistency protocols. The aim is to offer data availability while still preserving the consistency guarantees (application invariants). A large part of the operations on data can be designed to be asynchronous, synchronization will occur only when strictly necessary.

This thesis will focus mainly on data placement, data localization and indexation, but also on data consistency and fault-tolerance. Indeed, all these problems are tightly connected. The data is geo-distributed and replicated. The different copies of a data must be placed (and moved) judiciously, taking into account accesses (frequency and types) but also fault-tolerance criteria (in order to deal with correlated failures for example) privacy, etc. When the copies are placed according to specific criteria, it is not possible to use an implicit locating mechanism anymore, as it exists in distributed hash tables (DHT).  Thus, it becomes necessary to design indexing / localization mechanisms that provide efficient access to data. Moreover, in large-scale systems, nodes arrivals and departures are frequent. The work carried out in this thesis will take into account this volatility.

Keywords: Large scale Distributed Systems, Fault tolerance, Data placement

Director : Sebastien Monnet

Co-director : Flavien Vernier

Started : Octobre 2017

Doctoral School : SISEO « Sciences et Ingénierie des Systèmes, de l’Environnement et des Organisations » 


  • Algorithmic and Programming (PEIP, first year)
  • Graphs and Languages (IDU, first year)
  • Algorithmic Project (IDU, first year)
  • Conception and Programming Oriented Object (IAI, first year)