Managing server energy and operational costs in hosting centers

Abstract

The growing cost of tuning and managing computer systems is leading to out-sourcing of commercial services to hosting centers. These centers provision thousands of dense servers within a relatively small real-estate in order to host the applications/services of different customers who may have been assured by a service-level agreement (SLA). Power consumption of these servers is becoming a serious concern in the design and operation of the hosting centers. The effects of high power consumption manifest not only in the costs spent in designing effective cooling systems to ward off the generated heat, but in the cost of electricity consumption itself. It is crucial to deploy power management strategies in these hosting centers to lower these costs towards enhancing profitability. At the same time, techniques for power management that include shutting down these servers and/or modulating their operational speed, can impact the ability of the hosting center to meet SLAs. In addition, repeated on-off cycles can increase the wear-and-tear of server components, incurring costs for their procurement and replacement. This paper presents a formalism to this problem, and proposes three new online solution strategies based on steady state queuing analysis, feedback control theory, and a hybrid mechanism borrowing ideas from these two. Using real web server traces, we show that these solutions are more adaptive to workload behavior when performing server provisioning and speed control than earlier heuristics towards minimizing operational costs while meeting the SLAs.

The Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines

“…Chase et al [14], G. Chen et al [15], and Y. Chen et al [16] consider schemes for automatically provisioning resources in datacenters taking energy savings and application performance into account. Raghavendra et al [69] describe a comprehensive framework for power management in datacenters that coordinates hardware-level power capping with virtual machine dispatching mechanisms through the use of a control theory approach.…”

Power capping: a prelude to power shifting

“…Chen et al [11] also develop a controller to manage the response time in a server cluster. Although they both use control theory to manage power consumption, power is only used as a knob to control application-level service metrics.…”

Barely Alive Servers: Greener Datacenters Through Memory-Accessible, Low-Power States

“…Many papers have studied the combination of dynamic workload consolidation with server turn off [Chase et al(2001), Chen et al(2008), Chen et al(2005), Heath et al(2005), Pinheiro et al(2001), Rajamani and Lefurgy(2003)]. In this paper, we demonstrate how to make energy conservation in dynamic workload consolidation more practical through the creation of a family of active low-power server states.…”

“…As a result, previous works have proposed approaches for conserving energy in these services, such as consolidating workloads into a subset of servers and turning others off [Chase et al(2001), Chen et al(2005), Chen et al(2008), Pinheiro et al(2001)], and leveraging dynamic voltage and frequency scaling of the CPUs [Chen et al(2005), Elnozahy et al(2003), Fan et al(2007)]. …”