Enabling parallel and interactive distributed computing data analysis for the ALICE experiment

AliEn (ALICE Environment) is the production environment developed by the ALICE collaboration at CERN. It provides a set of Grid tools enabling the full offline computational work-flow of the experiment (simulation, reconstruction and data analysis) in a distributed and heterogeneous computing environment. In addition to the analysis on the Grid, ALICE users perform local interactive analysis using ROOT and the Parallel ROOT Facility (PROOF). PROOF enables physicists to analyse in parallel medium-sized (200-300 TB) data sets in a short time scale. The default installation of PROOF is on a static dedicated cluster, typically 200-300 cores. This well-proven approach is not devoid of limitations, more specifically for analysis of larger datasets or when the installation of a dedicated cluster is not possible. Using a new framework called Proof on Demand (PoD), PROOF can be used directly on Grid-enabled clusters, by dynamically assigning interactive nodes on user request. This thesis presents the PoD on AliEn project. The integration of Proof on Demand in the AliEn framework provides private dynamic PROOF clusters as a Grid service. This functionality is transparent to the user who will submit interactive jobs to the AliEn system. The ROOT framework, among other things, is used by physicists to carry out the Monte Carlo Simulation of the detector. The engineers working on the mechanical design of the detector need to collaborate with the physicists. However, the softwares used by the engineers are not compatible with ROOT. This thesis describes a second result obtained during this PhD project: the implementation of the TGeoCad Interface that allows the conversion of ROOT geometries to STEP format, compatible with CAD systems. The interface provides an important communication and collaboration tool between physicists and engineers, dealing with the simulation and the design of the detector geometry.