Analysis of database performances.

My work in this area was mainly centered about evaluating the sizes of "derived" relations in a relational database framework, in the hope ofunderstanding the performance of query optimisers. Interestingly, it turned out that these sizes follow a Gaussian distribution under a wide spectrum of assumptions, thus lending theoretical fundation to the empirical choices made by query optimisers, who limited themselves most of the time to the average value.

More recently, we also analyzed the occupancy rate of external nodes in B-trees, through a Polya urn model. Surprizingly, we were able to show that there is a phase transition and that "small" and "large" sizes of nodes give different asymptotic behaviours.

List of publications