—Server consolidation based on virtualization is a key ingredient for improving power efficiency and resource utilization in cloud computing infrastructures. However, to provide satisfactory performance in such scenarios under changing application workloads, dynamic management of the consolidated resource pool is critical. Unfortunately, this type of management is also challenging in cloud platforms because of the inherent tradeoffs between power and performance, and between the cost of an adaptation and its benefit. In this paper, we present Mistral, a holistic controller framework that optimizes power consumption, performance benefit, and the transient costs incurred by various adaptations and the controller itself to maximize overall utility. Mistral can handle multiple distributed applications and largescale infrastructures through a multi-level adaptation hierarchy and scalable optimization algorithm. Through extensive experiments, we show that our approach outstrips other st...
Gueyoung Jung, Matti A. Hiltunen, Kaustubh R. Josh