Computational grids clusters, provide powerful computing resources for executing applications of large scale. In Grid (clusters) usually several applications run simultaneously.The originality of Grid’5000 is that each application has characterized by its own specific requirement such as operating system (OS) or library components. Deploying the adequate OS needs to reboot the processors on which the application is executed. It is time-consuming and moreover frequent reboots may damage machines. In this work we investigate how to minimize the number of deployments, while keeping the running time as short as possible. We present the multiprocssors scheduling with d´eployement problem and provides a lis scheduling algorithm. The analysis details are presented in the worst case performence of the algorithm. 1 Because