Climate change is now widely documented and no longer a matter of debate in the scientific community. The origins of this disruption and its consequences for the earth's habitability are the subject of much research, and are discussed at various conferences and gatherings, such as the COPs. In 2015, the COP21 held in Paris led to the signing of the Paris Agreement, in which the various states agreed to significantly reduce global greenhouse gas (GHG) emissions with the aim of limiting global warming to 2°C this century, while continuing the action taken to limit it further to 1.5°C. If we are to meet the Paris agreements, global GHG emissions must fall considerably in all sectors, ICT being no exception. GHG emissions in the digital sector are divided between user terminals, communication networks, and datacenters (DCs). Datacenters account for a significant proportion of emissions, partly due to the manufacture of equipment and partly due to the large amount of electricity they consume. Growing demand for digital services has led to the democratization of hyperscale DCs, as well as a host of challenges for cloud operators. Among these challenges, the efficient management of computing and storage resources and the right sizing of infrastructures appear as two levers for limiting and reducing GHG emissions in DCs. Various techniques derived from virtualization have been studied as catalysts for these reduction levers, including elasticity and consolidation. These methods appear to be technical responses to the question of infrastructure oversizing, which may be caused by misguided user behavior. Today, cloud users often tend to overestimate their needs, biased by the perception of infinite resources offered by DC operators. But just as they are part of the problem, cloud infrastructure users can be part of the solution when it comes to limiting the impacts of these infrastructures. In this poster, we study how users equipped with a technical tool for automatically shutting down virtual machines can reduce the power consumption of a virtualization infrastructure. Part of the study is dedicated to profiling users, their understanding of the environmental impacts of DCs, and their knowledge of the infrastructure through a questionnaire. 19 interviews were also carried out to identify the VM profiles present in the infrastructure and to understand the trade-offs made by our users concerning whether or not to switch off their VMs. By coupling VM shutdown with a simple consolidation algorithm, we demonstrate in the poster that we can achieve a 30% reduction in energy consumption. This reduction in power consumption, generally made possible solely by the use of consolidation techniques, is amplified by 7% here through the involvement of users. Our study also highlights that it is not necessary to have a good understanding of the environmental impacts of DCs in order to do one's part. A certain knowledge of the infrastructure, combined with a limited amount of time, can significantly enhance the potential of technical solutions.