Cloud Computing Datacenters represent a power-intensive industry with well-known economical and ecological challenges. This work focusses on Virtual Machine Placement (VMP) problems as a valid alternative to address mentioned challenges. An experimental evaluation of 36 VMP optimization algorithms for power consumption minimization is presented. Algorithms were evaluated under uncertainty of 4 different dynamic parameters, considering 400 experimental scenarios and taking into account an average objective function cost as evaluation criterion. Experimental results indicate that two-phase algorithms considering prediction-based VMPr Triggering and update-based VMPr Recovering methods are best suited for power consumption minimization.CONACYT – Consejo Nacional de Ciencia y TecnologíaPROCIENCIA