Michel Raynal obtained bachelor degrees both in literature and science. He received his PhD from University of Rennes in 1975, and his “Doctorat d’état” in 1981. During the period 1981-1984 he was a professor in a telecommunications engineer school where he created and managed the informatics department. In 1984 he moved to the university of Rennes, and in 1985 he founded a research group entirely devoted to Distributed Algorithms. Michel Raynal has been an associate member of the editorial board of international journals, including the Journal of Parallel and Distributed Computing, IEEE Transactions on Computers, and IEEE Transactions of parallel and Distributed Systems, among others.
Research areas and scientific interests
Michel Raynal’s research contributions concern mainly concurrent and distributed computing, and more specifically: causality, distributed synchronization, fault-tolerance, distributed agreement and distributed computability. His first book is recognized as one of the very first books entirely devoted to distributed algorithms. On the synchronization side, with Jean-Michel Hélary and Achour Mostéfaoui, Michel Raynal designed a very simple generic message-passingmutual exclusionalgorithm from which can be derived plenty of token and tree-based mutex algorithms. On the causality side, with co-workers he produced a very simple algorithm for causal message delivery, and an optimal vector-clock-based distributed checkpointing algorithms, which established the theoretical foundations of distributed checkpointing, and the so-called communication-based snapshot. He also introduced the notion of virtual precedence. Together with V. Garg, he introduced the concept of “normality” which extends the well-known linearizability consistency condition to the case where objects have polyadic operations. On the agreement side, Michel Raynal produced several algorithms for asynchronous message-passing systems which solve consensus in the presence of crash failures or process Byzantine failures. This last algorithm is an incredibly simple randomized algorithm that is optimal with respect to both time and message complexities. With Mostéfaoui and Rajsbaum, Michel Raynal also introduced a new approach to solve consensus called “condition-based”. This approach brought to light a very strong connection between error-correcting codes and distributed agreement problems. Michel Raynal also designed distributed algorithms for other agreement problems. Recently, Armando Castaneda, Sergio Rajsbaum, and Michel Raynal introduced the notion of “interval linearizability” which is the first notion that allows us to unify in a single framework the notions of “concurrent objects” and “distributed tasks”. On the computability side, Steiner, Taubenfeld, and Raynal addressed universal constructions that allow x out of k distributed state machines to progress in the presence of asynchrony and any number of process crashes. Recently, from an initial idea proposed by Taubenfeld, Michel Raynal became interested in algorithms suited to anonymous memories.
Awards and Honours
2010: Senior Member, Institut Universitaire de France