Green Performance Indicators for Energy Aware IT Systems: Survey and Assessment

Green computing provides techniques to reduce the wastage of energy, making it very critical to the development of IT systems due to the increasing power and energy needed now-a-days to run data centers. Green metrics play a vital role in green computing since they are criteria for evaluating the green performance of the large scale IT systems. Green computing metrics need to be defined to measure power costs and energy consumption. Measuring the amount of energy consumed by IT systems is a direct way of quantizing the amount of energy wasted or used efficiently by data centers. Energy awareness in applications and data centers can be obtained and calculated through green metrics such as the Green Performance Indicators (GPIs). The GPIs are classified into four classes: IT Resource Usage GPIs that compute resource usage, the Application Lifecycle KPIs that define efforts required to develop or redesign applications and reconfigure IT-infrastructure, the Energy Impact GPIs that represent the environmental impact of data centers, and the Organizational GPIs that describe organizational factors. Dividing the GPIs into four classes is an approach accepted in the EU Project GAMES. Many metrics have been identified by many associations, but there still yet to exist a framework in which a set of metrics are defined and used by applications and data centers to measure energy efficiency. This paper is a survey that sums up all the known metrics defined by almost all associations, and explicitly states their different units, scopes, and most importantly compares between Journal of Green Engineering, Vol. 3, 33–69. c © 2012 River Publishers. All rights reserved. 34 S.S. Mahmoud and I. Ahmad them to define similarities and replacements that may exist. Exact similarities may help eliminate metrics whose measurements can be accomplished by another metric. The gathered GPIs are defined into the four main classes mentioned above and frameworks built based on the similarities found in each class, are developed. We also define a framework for the relationship between green computing and GPIs to illustrate how GPIs are incorporated in the process. Since GPIs measure different factors related to energy consumption, they may have different units. We furthermore compare and critique different approaches to unify the various units of the different metrics, and implement one of the techniques on most of the metrics mentioned below with some improvements. We introduce the idea of correlations which can measure the amount of similarities found between the metrics to reduce the number of metrics to unify.