Green Evaluation Metrics and Software Tool for Data Center



Similar documents
Heat Recovery from Data Centres Conference Designing Energy Efficient Data Centres

Green IT Promotion Council, Japan

Data Center Efficiency Metrics: mpue, Partial PUE, ERE, DCcE

Harmonizing Global Metrics for Data Center Energy Efficiency

The Green Grid, Metrics and DCMM. Presented by: Jay Taylor

ISSN :

Auditing a data centre but to what standard?

Energy Savings in the Data Center Starts with Power Monitoring

Managing Power Usage with Energy Efficiency Metrics: The Available Me...

Identifying Energy Saving Opportunities in the Data Center

DC Pro Assessment Tools. November 14, 2010

EU Code of Conduct for Data Centre Efficiency

Greener Pastures for Your Data Center

New Data Center Energy Efficiency Evaluation Index DPPE (Datacenter Performance per Energy) Measurement Guidelines (Ver 2.05)

DPPE: Holistic Framework for Data Center Energy. Efficiency. - KPIs for Infrastructure, IT Equipment, Operation (and Renewable Energy) -

Selecting an Industry-Standard Metric for Data Center Efficiency

Global Green Data Center Best Practices In Action

Recommendations for Measuring and Reporting Overall Data Center Efficiency

Measuring Energy Efficiency in a Data Center

DATA CENTER. Enhancing the Energy Efficiency and Use of Green Energy in Data Centers

How Does Your Data Center Measure Up? Energy Efficiency Metrics and Benchmarks for Data Center Infrastructure Systems

Greening The Data Center

ERE: A METRIC FOR MEASURING THE BENEFIT OF REUSE ENERGY FROM A DATA CENTER

Data Center Industry Leaders Reach Agreement on Guiding Principles for Energy Efficiency Metrics

DATA CENTERS: AN ACCOUNT OF FEDERAL INITIATIVES AND DATA CENTER ASSESSMENTS FOR FACILITY MANAGERS. David Cosaboon, LEED AP O+M

Recommendations for Measuring and Reporting Overall Data Center Efficiency

Data Center Energy Use, Metrics and Rating Systems

Green Data Centers. Energy Efficiency Data Centers. André ROUYER Director of Standardization & Environment. The importance of standardization

THE GREEN GRID DATA CENTER POWER EFFICIENCY METRICS: PUE AND DCiE

Data Center Energy Efficiency Looking Beyond PUE

DATA CENTERS BEST IN CLASS

DataCenter 2020: first results for energy-optimization at existing data centers

Power Usage Effectiveness (PUE) & Data Center Infrastructure Efficiency (DCiE) Progress

Measuring Power in your Data Center: The Roadmap to your PUE and Carbon Footprint

Guideline for Water and Energy Considerations During Federal Data Center Consolidations

General Recommendations for a Federal Data Center Energy Management Dashboard Display

Metrics for Data Centre Efficiency

A Paradigm Shift in Data Center Sustainability Going Beyond Aisle Containment and Economization

Building a data center. C R Srinivasan Tata Communications

GREEN GRID DATA CENTER POWER EFFICIENCY METRICS: PUE AND DCIE

Energy Efficiency Monitoring in Data Centers: Case Study at International Hellenic University. George Koutitas, Prof. L. Tassiulas, Prof. I.

Operating Sustainable Facilities

A Market Transformation Programme for Improving Energy Efficiency in Data Centres

Understanding Power Usage Effectiveness (PUE) & Data Center Infrastructure Management (DCIM)

AbSTRACT INTRODUCTION background

Server Consolidation for SAP ERP on IBM ex5 enterprise systems with Intel Xeon Processors:

Server Migration from UNIX/RISC to Red Hat Enterprise Linux on Intel Xeon Processors:

Potpourri Track Green IT. Colin Martin Solutions Architect, Cisco Systems Inc.

Datacenter Efficiency

Green ICT: Consistent Actions to Reduce Energy Consumption

Auditing a Data Centre But to What Standard?

How green is your data center?

ENERGY STAR for Data Centers

StruxureWare. How to reconcile efficiency and availability in a virtualized data center. for data centers

Page 1. White Paper, December 2012 Gerry Conway. How IT-CMF can increase the Energy Efficiency of Data Centres

GLOBALWORKPLACESOLUTIONS-ENERGY SERVICES. Reduce energy costs and greenhouse gas emissions across your portfolio

STATEMENT OF. Dr. David McClure Associate Administrator Office of Citizen Services and Innovative Technologies General Services Administration

Data Centers: Definitions, Concepts and Concerns

Sustainable Performance for Data Centres. DCEM-Plus

Oracle Platform as a Service (PaaS) FAQ

Green data center: how green can we perform?

Code of Conduct on Data Centre Energy Efficiency. Endorser Guidelines and Registration Form. Version 3.0.0

KWE Basic Philosophy. As of March 2010, KWE is ISO14001 certified at 6 locations in Japan and 8 locations overseas

Prudential plc. Basis of Reporting: GHG emissions data and other environmental metrics.

Energy Management Solutions for a Better and Greener Government

How to Earn the LEED Green Power Credit

The Mission Critical Data Center Understand Complexity Improve Performance

Energy and Sustainability-- Green IT in a holistic approach to slashing energy use, emissions and the impact to the environment

GHG Protocol Product Life Cycle Accounting and Reporting Standard ICT Sector Guidance. Chapter 8: 8 Guide for assessing GHG emissions of Data Centers

ENERGY EFFICIENT DATA CENTRES AND STORAGE. Peter James University of Bradford

Green Cloud Computing: Case Study Sri Lanka & Pakistan

Transcription:

Green Evaluation Metrics and Software Tool for Data Center Wong Hui Shin Faculty of Computing and Informatics, Multimedia University, Cyberjaya, Selangor, Malaysia. hswong@mmu.edu.my Saravanan Muthaiyah Faculty of Management, Multimedia University, Cyberjaya, Selangor, Malaysia. saravanan.muthaiyah@mmu. edu.my Murali Raman Faculty of Management, Multimedia University, Cyberjaya, Selangor, Malaysia. murali.raman@mmu.edu.my Abstract: - United States of America has the largest number of data centers while China and European Union are catching up rapidly. Data centers today do not have a standard measure to evaluate their level of green compliance or GREENNESS. The report Congress on Server and Data Center Energy Efficiency Public Law 109-431 made by the U.S. Environmental Protection Agency Energy Star Program on August 2, 2007 urged the collaboration between government and industry to develop software tools for Data Center Energy effciency. Green evaluation metrics and software tool for data center are studied, planned and designed. In this study, we explore variables based on the 3Rs principle: Reusing, Reducing and Recycling and propose a set of metrics that combines all aspects of 3R from tier-1 to tier-4 data centers. A new metric evaluation for data centers is necessary for data centers to self-evaluate their level of greenness based on 3R so that those who are GREEN can then be given subsidies and vice versa. This paper also illustrates the architecture and the framework of the green evaluation software tool for data center and it is also known as GreenEval. It is a software tool to aim for benchmarking data centers using United States Environmental Protection Agency (EPA) s online tool concept and objective. It aims to justify the data center green evaluation from quantitative perspective and operating cost perspective. The green evaluation software tool is also focus on assessing the performance implication of Reusing, Recycling and Reducing during periodic evaluation process. Key-Words: - green evaluation metrics, green evaluation software tool, reducing, recycling and reusing. 1 Introduction In 2007, U.S. Environmental Protection Agency (EPA) attracted the global attention towards data centers environmental impacts by its Data Center Report to Congress. Afterwards, a range of different industry leaders such as Cisco and Oracle, as well as the well-known environmental stewardship programs like Energy Star and LEED have focused their efforts on data centers GREENNESS. One of the important milestones in this issue was introducing metrics for measuring data centers GREENNESS. In 2007, Green Grid Institution introduced two metrics, which are the most common data center efficiency metrics nowadays: PUE and DCiE [1]. Later on, many other researchers and industry groups proposed new metrics mostly aimed for reducing power consumption and improving cooling efficiency, whereas a comprehensive green metric should include other aspects of 3Rs such as waste disposal and recycling. Only recently limited recycling and reusing metrics were offered, however, there is still no universal standard of metrics available for evaluating the overall green compliance of a data center [2][3][4]. Since 2011, a few of these data centers like NTT and CSF have started using green metrics [5], but there is still a long way into clarifying, establishment and monitoring green data center metrics in Malaysia and worldwide [6]. According to the report Congress on Server and Data Center Energy Efficiency Public Law 109-431 made by the U.S. Environmental Protection Agency Energy Star Program on August 2, 2007, the federal government should issue a challenge to private sector Chief Executive Officers (CEOs) to conduct U.S. Department of Energy (DOE) Save Energy Now energy efficiency, implement improvements and report energy performance of their data centers [7]. It states the urgency for joint software tool development between the government and the industry to do the assessments with a list of protocol ISBN: 978-1-61804-147-0 25

and tools. It also states the urgency to have joint development between federal government and the industry to develop improved tools, such as energy aware total cost of ownership models and life-cycle risk models, for management of energy in data centers. A development of energy aware TCO methods, data and tools and the development of tools and standards for chargeback systems, and document best practices are needed. To date, DC Pro Profiling Tool and Air Management Tool were developed after the urge. DC Profiling Tool is to identify the potential savings and to reduce environmental emissions associated with data center energy use. The Air Management Tool in data centers is important both for energy and thermal management. The Electronic Product Environmental Assessment Tool (EPEAT) provides a registration label for products that meet certain environmental and energy attributes. The available software development tools include life-cycle risk models and total cost of ownership models that incorporate energy costs, for management of energy in data centers. Department of Environment (DOE) s Save Energy Now program provides tools, training, best practice information and energy assessments directly to data centers. There are more than 3,200 buildings have earned the ENERGY STAR rating and more than 26,000 have used EPA s online tool to benchmark energy performance on the 1-to-100 scale in United States of America. There is no current benchmark for data centers. The EPA s online tool is for the commercial buildings. Thus, it is an urgency to build a tool to benchmark data centers using the EPA s on-line tool concept and objective. The purpose of this study is to design a comprehensive set of metrics, including all of the 3Rs principle to evaluate the data centers total green level and later to transform the set of metrics into a unique 3 color-coding dashboard (Green=Good, Yellow=Average, Red=below industry standards), that is going to be the first comprehensive diagnostic tool in the world for data centers to assess and monitor their overall green compliance. 2 Green Evaluation Software Tool: GreenEval This research aims to introduce a green evaluation software tool, GreenEval that will result in a set of metrics to evaluate data centers based on the 3R principle i.e. Reduce, Reuse, and Recycle. Many evaluation tools are currently available such PUE, DCE, and SI-POM but very few assess the data center in terms of 3Rs (Miller, 2010), (Padmanabhan, 2011). In this project, we apply Analysis phase and Design phase in Software Development Life Cycle (SDLC) to find out the input, process and output of the software tools and the standards for continual improvement in data center energy performance; firstly, we need to gather all the input parameters with wide range of available metrics, secondly, we calculate the scores for the input based on the existing mathematical formulas for the metrics, later, we consolidate and normalize the output scores for 3Rs and eventually, we identify the architecture of the green evaluation software tool. The output of this project is evaluation software tool architecture with 3 color-coding greenness performance scale (Green=Good, Yellow=Average, Red=below industry standards) for assessing the data centers greenness respecting all 3Rs of waste management. The greenness performance scale answers the question: how green is my data center compared to others nationwide? Is it good, average or below industry standard. Figure 1: GreenEval s Evaluation Mechanism Figure 1 illustrates input, process and output of the software tool. It shows the 10 top level measures (GEC, ERF, ERE, MRE, PUE, ITEU, ITEE, WUE, CUE and MRE) and 3 underlying measures (Score X, Score Y and Score Z). Later, these 3 scores will be normalized by a greenness normalization formula as stated in the later section. The final output will show a value in 3 color-coding greenness performance scale: Green, Yellow or Red. ISBN: 978-1-61804-147-0 26

Note: ERE (Energy Reuse Effectiveness), MRE (Material Reuse Effectiveness), GEC (Green Energy Coefficient), ERF (Energy Reuse Factor), PUE (Power Usage Efficiency), ITEU (IT Equipment Utilization), ITEE (IT Equipment Energy Efficiency), WUE (Water Usage Effectiveness), CUE (Carbon Usage Effectiveness) and MRE (Material Reuse Effectiveness) We list down the working formulas for the mentioned top 10 level measures. ERE = (Cooling + Power + Lighting + IT-Reuse)/ IT MRE = (Total In-Bound Material Outbound Product & Service in Mass [lbs/ Kg]) / (Total Recycled/ Reclaimed/ Reused*Material measured in Mass [lbs/ Kg]) ERF = Reuse Energy/ Total Energy PUE = Total Energy/ IT Energy ITEU = IT Equipment Utilization ITEE = IT Equipment Work Capacity/ IT Equipment Energy WUE = Annual Water Usage/ IT Equipment Energy CUE = Total CO2 emissions caused by the Total Data Center Energy/ IT Equipment Energy GreenEval will be developed in Java Platform. 3 EPA Online Tool Concept and Objective EPA uses Portfolio Manager with energy performance rating system to let citizens to rate their building energy performance. It benchmarks with other buildings in the country with the U.S. Department of Energy s Commercial Building Energy Consumption Survey (CBECS) conducted every four years. The survey forms the basis for most ENERGY STAR energy performance scales. Portfolio Manager is an interactive energy management tool that allows the building owners or tenants to track and assess energy and water consumption across entire portfolio of buildings in a secure online environment. It set investment priorities, identify under-performing buildings, verify efficiency improvements, and receive EPA recognition for superior energy performance on a scale of 1-100 relative to similar buildings. An ENERGY STAR label will be awarded when the buildings rating 75 or greater. EPA Online Tools for Buildings Benchmarking [8] ENERGY STAR Energy Performance Scale of 1-100 with building rating 75 and greater will be awarded ENERGY STAR label. Energy s Commercial Building Energy Consumption Survey (CBECS) conducted every four years is the basis of the ENERGY STAR Energy Performance Scale. Online To determine the score rating from 1-100, it uses the statistic model to predict the building s energy use, given its size, location, hours of operation, and other relevant characteristics. Then, calculate the building s energy efficiency ratio and determine where this ration places the building on the 1-100 scale. Green Evaluation Software Tool for Data Center: GreenEval 3 color-coding Greenness Performance Scale of 1-100 with green rating 1-33.33 is Red = below industry standards, 33.34-66.66 is Yellow = Average, 66.67-100 is Green = Good. Data Center Energy Performance and Consumption Survey (DCEPCS) conducted first ever in Malaysia in year 2012 is the basis of 3 color-coding Greenness Performance Scale. Online To determine the score rating from 1-100, it uses the top 10 level measures to calculate the input for the 3 underlying measures. Later, these 3 scores of the 3 underlying measures will be normalized by a greenness normalization formula as stated Section 5. The final output will show a value in 3 color-coding greenness performance scale: Green, Yellow or Red with the respective score. Table 1: The Comparison of EPA Online Tools for Buildings Benchmarking and Green Evaluation Software Tool for Data Center: GreenEval ISBN: 978-1-61804-147-0 27

4 Research Methodology The methodology of this study includes following methods: 1) Delphi method: a group discussion with 100 field experts from each data center in order to get the professional point of view. 2) Interviews: interviewing data center managers will be the next step to go more in depth of what have been already achieved from the Delphi discussion. 3) Data Center Energy Performance and Consumption Survey (DCEPCS) 2012: data center managers will fill in the survey according to questions listed to access the 10 top level measures (GEC, ERF, ERE, MRE, PUE, ITEU, ITEE, WUE, CUE and MRE) 4) Statistic Analysis on the DCEPCS outputs. This project uses Software Development Life Cycle (SDLC) to complete the evaluation tool s development. The process methodology is divided into 3 stages as shown below i.e. (stage 1 for reuse, stage 2 for reduce and stage 3- for recycling). The Emerson and Leeds metrics are reviewed for suitable metrics to be applied in each stage. Then, the metric scores from each stage are combined and they are normalized to a standard scale. It starts with the exploratory of the evaluation tool in Analysis and Design phase in Software Development Life Cycle (SDLC). The Analysis and Design phase involves the reviewing of the previous works have been done related to Reducing, Reusing and Recycling. We follow the process chart to work on the Analysis and Design phase of engine development for the green evaluation software tool. 5 GreenEval s Evaluation Mechanism GreenEval s Evaluation Mechanism works on quantitative perspective and operating cost perspective. From the quantitative perspective, the mechanism starts with 10 top level measures (GEC, ERF, ERE, MRE, PUE, ITEU, ITEE, WUE, CUE and MRE) and 3 underlying measures (Score X, Score Y and Score Z). These 3 scores will be normalized by a greenness normalization formula. The final output will show a value in 3 color-coding greenness performance scale: Green, Yellow or Red. From the operating cost perspective, the mechanism works with the relativity of number of rest hours for equipments in data center operations. The equipments schedules for maintenance will directly influence the operating cost for a data center. 5.1 Quantitative Perspective In our study, we categorize 3 types of metrics for data centers: Reducing Metrics, Reusing Metrics and Recycling Metrics. These metrics are known as the top 10 level measures in green evaluation software tool, GreenEval. We obtain the 3 underlying measures (Score X, Score Y and Score Z) by multiplying the respective top 10 level measures to a rating of 100% for the relative categories. The Score X, Score Y and Score Z are obtained using the formulas below with the top 10 measures that already converted to a rating of 100%. Formula 1: Score X = GEC(%)*ERF(%)*ERE(%)*MRE(%) Formula 2: Score Y = PUE(%)*ITEU(%)*ITEE(%)*WUE(%)* CUE Formula 3: Score Z = MRE(%) The greenness normalization formula to obtain the value for the rating of the data center greenness, A is defined as Score X(%)*Score Y(%)*Score Z(%). Formula 4: A = Score X(%)*Score Y(%)*Score Z(%) A is a rating from 1-100 with a 3 color-coding greenness performance scale: Red (1-33.33%), Yellow (33.34-66.66%) or Green (66.67-100%). 5.1.1 Reducing Metric Energy (PUE Power Usage Effectiveness, ITEU - IT Equipment Utilization, ITEE - IT Equipment Energy Efficiency) ISBN: 978-1-61804-147-0 28

Water (WUE Water Usage Effectiveness) Carbon Dioxide (CUE Carbon Usage Effectiveness) 5.1.2 Reusing Metric Energy (ERE - Energy Reuse Effectiveness, GEC - Green Energy Coefficient, ERF - Energy Reuse Factor) Material (MRE Material Reuse Effectiveness) 5.1.3 Recycling Metric Material (MRR Material Recycling Ratio) 5.2 Operating Cost Perspective We would consider the data center equipments schedules for maintenance will directly influence the operating cost for a data center. We discuss about number of rest hours per equipment per year. The convention goes with an example of 10 equipments in a data center will operate 365 days, 24 hours a day with a 1 KWatt of electricity is 0.20 cents per KWatt per hour. The total of money spent per year for the setup above is 1752 * the number of KWatt used. With a equipments maintenance schedule of 1 equipment will spend 1 week (7*24 hours) for maintenance with a plan to allow 1 equipment to rest for each consecutive month. We will spend a total of 1416 * the number of KWatt instead of 1752 * the number of KWatt. We will save 336 * the number of KWatt. In short, the equipments maintenance schedule is very important. We will plan to work on an optimal schedule for generic solution for data center operation. 6 Performance Implication Besides, we focus on assessing the performance implication of Reusing, Recycling and Reducing metrics, the top 10 measures and the normalized Score X, Score Y and Score Z during periodic evaluation process. We justify the performance implication using Geometric mean, Net Present Value Analysis and Sensitivity Analysis to ensure that our final output from the green evaluation software tool, GreenEval is a trustable value after a series of benchmarking test from the survey, questionnaire and interviews. 6.1 Geometric Mean Geometric Mean is a type of mean or average which indicates the central tendency of typical value of a set of numbers. The Geometric Mean compares different items, finding a single figure of merit for these items when each item has multiple properties that have different numeric ranges. The top 10 measures have different measurements, different range and different comparison ratio. For example, Power Usage Effectiveness (PUE) is a ratio of total amount of power used by a computer data center facility to the power delivered to computing equipment. The ideal PUE is 1.0. We map the 0-1.0 rating scale for PUE to 1-100 rating scale for PUE(%) in Formula 2 for Score Y. IT Equipment Utilization (ITEU) is a metric to promote reduction in energy consumption by improving utilization rate of IT equipment and reduction of surplus equipment investment. It is also a metric to evaluate efforts in design and operation of IT equipment in a Data Center. The ideal rate is 100%. In this case, we need not to convert or normalise the ITEU. IT Equipment Energy Efficiency (ITEE) is a ratio of the total rated capacity of IT equipment and the total rated energy consumption of IT equipment. In this case we need to map rating of ITEE to ITEE(%), 1-100 rating scale. Also, we work on the 3 color-coding greenness performance scale: Green, Yellow or Red. The accuracy at the point of 33.33%, 66.66% and 99.99% has 0.01 error tolerances. We need to look for a better way to find a more accuracy calculation for the final output. 6.1 Net Present Value Analysis Due to the nature of the top 10 level measures, the cost of error is exceptionally high. The higher quality of the accuracy for the measurement may far more important in a proper evaluation of effectiveness than performance or availability. To obtain more accurate measurements and values, an expected net present value analysis is needed. The value of benchmarking is base on the survey values from a wide variety of data center inputs and outputs for various metrics. The net present value ISBN: 978-1-61804-147-0 29

analysis mainly deals with the GreenEval Evaluation Mechanism from the operating cost perspective. It is for a more accurately forecast for cost saving in a data center. 6.1 Sensitivity Analysis All equipments in data centers have a life span. The performance of the equipments will decrease in a long period of time. The sensitivity analysis is important to identify the border lines for the points at 33.33%, 66.66% and 99.99%. A series of testing should be encountered for a more accurate data output from the green evaluation software tool, GreenEval. Manage. Environ. Qual.: International Journal, Volume 20, page 712-731, 2009. [7] United States Environmental Protection Agency (EPA) ENERGY STAR Program: Report to Congress on Server and Data Center Energy Efficiency Public Law 109-431, 2007. [8] United States Environmental Protection Agency (EPA) ENERGY STAR Program: Understanding EPA s Energy Star Energy Performance Scale, 2007. 7 Conclusion Due to the needs and urgency of developing software tool to enhance any performance perspective of data center, we propose GreenEval, a green evaluation software tool with concise evaluation mechanism. The software goes through performance implication for normalization; 1) Geometric Mean, 2) Net Present Value Analysis and 3) Sensitivity Analysis. We will work intensively on the equipments maintenance schedule, so that we can save some costs for data center operation. We are looking forward for a government grant to develop the green evaluation software tool, GreenEval. References: [1] A. Rawson, J. Plueger, T. Cader, The Gren Grid Data Center Power Efficiency Metrics: PUE and DCiE, The Green Grid: Metrics and Measurements White Paper, 2007. [2] B. Tschudi, O. Vangeet, J. Cooley, D. Azevedo, ERE: A Metric for Measuring the Benefit of Reuse Energy from a Data Center, The Green Grid: Metrics and Measurements White Paper, 2010. [3] Emerson Annual Report, 2010. [4] United States Environmental Protection Agency (EPA) s 2007 Report on the Environment: Science Report (SAB Review Draft) [5] Business Times Magazine, 2012. [6] T. Daim, J. Justice, M. Krampits, M. Letts, G. Subramanian and M. Thirumalai, Data Center Metrics: An Energy Efficiency Model for Information Technology Managers, ISBN: 978-1-61804-147-0 30