While Grid Computing is becoming more mature and has demonstrated its usefulness at large scale, a new approach of large scale distributed computing is proposed: Cloud Computing. Like for Grid, P2P and other large scale distributed systems, we believe that research in Cloud Computing needs real life experimental platforms. Indeed, experimental platforms of large scale distributed systems (PlanetLab, Grid'5000, DAS) have demonstrated strong merits as exploration tools for computer science 1) by enabling experiments of complex systems at large scale, 2) by providing high quality experimental results, complementary to those obtained theoretically or by simulation, 3) by establishing reference testbeds accepted by the community and 4) by boosting the research in these domains. While Cloud Computing is emerging as a significant research area, it is time to understand, from what we learned with existing experimental platforms: a) what are the essential concepts of a real life experimental platform like Grid'5000 and what are the main difficulties to run it and 2) what is needed to conduct Cloud Computing experiments on a large scale platform like Grid'5000. This talk starts by presenting the latest evolution of some of the main Grid Initiatives in Europe: CoreGrid, XteemOS, DEISA, EGEE, etc. As a matter of fact, most of these projects have used experimental platforms like Grid'5000 or real scale prototypes for their research. In a second part, we recall the fundamental mechanisms of Grid'5000 that make it a highly versatile and flexible experimental infrastructure. The presentation of selected results will illustrate these properties. Then, we introduce the evolution of Grid'5000, through the ALADDIN project supported by INRIA. ALADDIN is expected to widen the research scope of Grid'5000. In the last part of the talk, we present some of the identified research issues in Cloud Computing and describe how Grid'5000 could be used for large scale experiments in this domain.