Performance Optimization within a Power Constraint using Precision Measurement and Feedback Control

Power consumption has arguably become the most important design consideration for modern, high-density servers, but current power management implementations have not evolved beyond primitive responses to thermal emergencies and do not manage to varying power and cooling constraints. We present a technique that manages the peak power consumption of a high-density server by implementing a feedback controller that uses precise, system-level power measurement to periodically select the highest performance state while keeping the system within a fixed power constraint. A control theoretic methodology is applied to systematically design this control loop with analytic assurances of system stability and controller performance, despite unpredictable workloads and running environments. This technique is particularly valuable when applied to servers with multiple power supplies, where a partial failure of the power supply subsystem can result in a loss of performance in order to meet a lower power constraint. Conventional servers use simple open-loop policies to set a safe performance level in order to limit peak power consumption. We show that closedloop control can provide a more graceful degradation of service under these conditions and test this technique on an IBM BladeCenter HS20 server. Experimental results demonstrate that closed-loop control provides up to 82% higher application performance compared to open-loop control and up to 17% higher performance compared to a widely used ad-hoc technique.