Research Project Information

Research Project Summary Information

Energy Conscious Scheduling of Linux servers(ST10946-1)

The Research Foundation of SUNY at Bing


The New York and United States data center industry is in the midst of a major growth period stimulated by increasing demand for data processing and storage. During the past five years, increasing demand for computer resources has led to significant growth in the number of data center servers, along with an estimated doubling in the energy used by these servers and the power and cooling infrastructure that supports them. The energy used by the nation’s servers and data centers is significant. It is estimated that this sector consumed about 61 billion kilowatt-hours (kWh) in 2006 (1.5 percent of total U.S. electricity consumption) for a total electricity cost of about $4.5 billion. Adding to all of this is the trend to overprovision data center capacities, and the use of overrated power supplies for the individual servers. Such over provisioning results in gross energy inefficiencies, as servers and power supplies are generally designed to give very high energy efficiencies only at or near peak loading levels.

Map Error

Project Description

The approach taken in this project is to address the lack of energy-efficiency in servers at the root by treating energy as a first-class resource in scheduling jobs to Linux servers, and by using a coordinated, multi-tiered approach to energy management. The market share and TCO benefits of Linux servers have been growing steadily for the past five years and this project thus addresses an important and dominant part of the data center market. The concepts developed in this effort can also be adapted for all types of servers in general. The project takes a holistic, all-software approach to implement an energy-conscious scheduling of the server requests that allocate jobs and an energy budget to server racks and server blades within a data center. Significant advantages are gained by learning about the energy and performance characteristics of repetitive server jobs to guide request scheduling and for choosing the most energy-efficient and performance-effective settings for these jobs. The proposed energy management policies also attempt to operate the active blade servers near their peak operating regions, which are also the operating range where they have the highest energy efficiency. Traditional energy management features (such as processor speed-power stepping, disk power state control) are also incorporated as part of our tiered solution for energy management. The technology developed in this effort can also be adopted for non-Linux servers.


Benefits will be quantified via a demonstration of a prototype implemented on a test bed in the Linux Technology Center at Binghamton University. Expected results are a 15 to 25% enhancement in overall server energy efficiency beyond what is provided by the current state-of-the-art.

Project Results


The Research Foundation of SUNY at Bing
P.O. Box 9
Albany, NY 12201

Principle Investigator

Kanad Ghose

Universities Involved


Project Type:

Product Demonstration

Technologies Types:

Building Systems
Data Centers/Information Technology

NYSERDA Contact Information

Bryan Berry


R&D - Innovation & Biz Dev

Contract Details

Start Date: 10/2/2009
Project Status: Active
Contract Number: ST10946-1

Last Updated: 6/16/2014