Leakage-Aware Cooling Management for Improving Server Energy Efficiency

Abstract

Abstract-The computational and cooling power demands of enterprise servers are increasing at an unsustainable rate. Understanding the relationship between computational power, temperature, leakage, and cooling power is crucial to enable energy-efficient operation at the server and data center levels. This paper develops empirical models to estimate the contributions of static and dynamic power consumption in enterprise servers for a wide range of workloads, and analyzes the interactions between temperature, leakage, and cooling power for various workload allocation policies. We propose a cooling management policy that minimizes the server energy consumption by setting the optimum fan speed during runtime. Our experimental results on a presently shipping enterprise server demonstrate that including leakage awareness in workload and cooling management provides additional energy savings without any impact on performance.

An experimental and theoretical investigation of the effects of supply air conditions on computational efficiency in data centers employing aisle containment

“…The results of the GSWT experiments show that, for the server investigated, computational efficiency is not significantly affected by T supply or p CH within the ranges investigated. This was unexpected, with results published by Zapater et al [27] demonstrating increasing server power consumption with increasing T supply for an unspecified enterprise server. Similarly, the authors of the present paper note that preliminary tests using a more power dense server have shown computational efficiency of the server to be impacted significantly by p CH and T supply .…”

“…Various system models have been described in the research literature which seek to predict the impact of design and operation variables on DC power consumption and cooling effectiveness [12], [18], [21]- [26]. In addition, Zapater et al [27] have investigated the impact of server fan control algorithms and supply temperature (T supply ) on computational efficiency of the server. Only Tatchell-Evans et al have used a system model to investigate the impact of bypass on E T [18].…”

“…This is where the processing of digital information by the DC is facilitated, and is also the greatest source of heat, due to electrical impedance in the increasingly densely populated microelectronics. These components are highly sensitive to the thermal environment and require constant cooling without which they would exceed their thermal envelopes, causing increased power consumption, failures and warranty breaches associated with exceeding temperature limits and loss of service [27].…”

Predictive Dynamic Thermal and Power Management for Heterogeneous Mobile Platforms

“…Most of the previous work addresses power and temperature separately, while we take into consideration their inter-dependencies using the power and thermal models. [51] considers leakage power and temperature dependence, using which the authors control the fan speed for cooling of data center servers, while we deal with mobile platforms where fan is not an option. Certain power simulators assume a constant ratio between leakage and dynamic power [28,48].…”

Beyond CPU: Considering Memory Power Consumption of Software

“…Several factors contribute to the CPU power consumption and globally it is possible to give the following formula (1) in order to describe the power consumed by the CPU: P CPU = P CPU,dynamic + P CPU,sc + P CPU,leak (1) where , represents dynamic power consumption, , corresponds to short-circuit power consumption and , , power loss due to transistor leakage currents and varies with the temperature (Zapater et al, 2015). The last two power are due to at the hardware manufacturing.…”