Computer scientists working on the design of hardware and software architectures and in particular on the design of distributed architectures (network, high performance computing, cloud, sensor networks, IoT, etc.) need to evaluate the relevance of their proposal at scale on a regular basis. Hence, their practice relies on frequent experimental evaluations, which leads to specific needs in term of experimental control. In this context, reproducing the work of other colleagues happens to be very difficult as it requires not only to have precise information about experimental conditions (software stack, external load, hardware type and configuration, etc.) but also to have a testbed allowing to recreate similar experimental conditions. A few experimental testbeds allowing for fine-grain experimental control have been built in the last decade (Grid5000, R2Lab, …). Such testbeds are mutualized and generally open-access, which makes it possible to compare in a fair and truthful way alternative approaches at large scale. In this webinar, Lucas Nussbaum will provide an overview of these projects and of their internals.