[en] Centralized computers have been replaced by 'client/server' distributed architectures which are in turn in competition with new distributed systems known as 'peer to peer'. These new technologies are widely spread, and trading, industry and the research world have understood the new goals involved and massively invest around these new technologies, named 'grid'. One of the fields is about calculating. This is the subject of the works presented here. At the Paris Orsay University, a synergy emerged between the Computing Science Laboratory (LRI) and the Linear Accelerator Laboratory (LAL) on grid infrastructure, opening new investigations fields for the first and new high computing perspective for the other. Works presented here are the results of this multi-discipline collaboration. They are based on XtremWeb, the LRI global computing platform. We first introduce a state of the art of the large scale distributed systems, its principles, its architecture based on services. We then introduce XtremWeb and detail modifications and improvements we had to specify and implement to achieve our goals. We present two different studies, first interconnecting grids in order to generalize resource sharing and secondly, be able to use legacy services on such platforms. We finally explain how a research community like the community of high energy cosmic radiation detection can gain access to these services and detail Monte Carlos and data analysis processes over the grids. (author)

[en] CELESTE is an atmospheric Cherenkov telescope based on the sampling method which makes use of the de-commissioned THEMIS solar electrical plant in the French Pyrenees. A large (2000 m²) mirror surface area from 40 independent heliostats followed by a secondary optic, a trigger system using analog summing techniques and signal digitization with 1 GHz flash ADCs make possible the detection of cosmic γ-rays down to 30 GeV. This paper provides a detailed technical description of the CELESTE installation