I am a Senior Researcher at the University of Fribourg, Switzerland at the eXascale Infolab. Previously, I spent some time in research groups at IBM Research (both at T.J. Watson and Almaden), and in the industry, at Google and MongoDB – and some start-ups. I got my Ph.D. in 2003 at Ecole Normale Superieure de Telecomunications at Paris (now Paris Tech) having done the research work connected to my Thesis both at INRIA/Rocquencourt and at New York University, under Dennis Shasha.
My research interests revolve around non-conventional database systems. I mean non-conventional in at least three different contexts.
Systems that support much larger data than usual. At the high-end of the size scale, the established techniques are pushed to their breaking point. I have first-hand experience with problems arising in datacenter-sized systems where dozens of thousands of machines work cooperative to manage databases with many trillions of objects. One of my goals is to keep pushing this size boundary.
Systems that explore closely couplings with hardware as a way to realize untapped performance. The challenge here is to harmonize the operations a database wants to do with those that the underlying hardware is good at. For that we look even beyond the CPU itself. We found that there are opportunities to better integrate database systems with both storage and networking stacks. On an ongoing work, we are integrating the compute capacity of high-speed switches – the programmable ones – into massively parallel databases, and are thus capable of processing portions of a query at the switches, while they are moving the tuples around.
Systems that support data models other than relational. For instance, data that comes in arrays and sequences in a given order. My Thesis described a way to manipulate such ordered data in a declarative way, similar to SQL. Queries over data coming from fields such as Finance and Biology turned out much easier to express. And, as a result, we uncovered some new opportunities for query optimization.