Working Towards A Smarter Data Center
Use the tools and metrics at your disposal to maximize data center efficiency.
The pursuit for a holistic metric that covers the complete data center performance is still ongoing. Data center efficiency encompasses electricity entering the data center and a fraction of that electricity being used to power vital IT equipment. Very much like comparing real and apparent power, savings are possible by improving airflow for cooling, adopting unique and free cooling methods, upgrading cooling equipment, and utilizing renewable energy all managed and controlled by an intelligent energy management system. Efforts are being made constantly to cut energy used by IT equipment by upgrading to lower power CPUs, using more efficient hard drives, use of virtualization technology, etc.
Power usage effectiveness (PUE) and data center infrastructure efficiency (DCiE) have been popular tools for benchmarking efficiency. This has led to the well-known “PUE Wars.” The Green Grid, author of both PUE and DCiE, didn't aim for these metrics to be used for comparing one data center to another. While this is a good starting point, it is important to understand that these metrics by themselves fall short in determining data center efficiency as they don’t include productivity.
Important questions to ask include:
• Are you getting the most ‘bang for buck’ from your servers and storage?
• Are you maximizing processing power?
• Are you retiring idle servers?
• Are you consolidating and virtualizing?
• Are you utilizing renewable energy as opposed to conventional energy sources?
• Are you utilizing wasted energy in a productive manner?
Answers to these questions can be quite revealing as to how you’re achieving true efficiencies. By looking beyond PUE, you’ll be able to manage other major aspects of power consumption which is a better way to compare one data center design and operations versus another.
Better ways of measuring, calculating, and improving PUE and DCiE are necessary. Many of us have seen simple PUE calculators on websites that compare your data center’s power consumption to the existing IT loads. PUE therefore is the measure of how efficiently a data center utilizes power, i.e., power used by computing equipment in contrast to cooling and other overhead. PUE fails to measure the useful IT work done by the data center and therefore is not satisfactory for representing the overall efforts for energy savings made by the data center designers and operators.
PUE= (Total Facility Power)
(IT Equipment Power)
DCiE= (IT Equipment Power)
(Total Facility Power)
In their report to the U.S. Congress, the U.S. Environmental Protection Agency (EPA) provided a set of efficiency improvement scenarios that predicted future data center energy consumption and associated PUE values to 2011. The EPA listed four categories of data center efficiency improvements, with increasing cost and complexity, summarized in Table 2.
Total facility power is made up of a number of data center equipment. Typical power consumption is provided by The Green Grid data center model in Figure 1.
Data Center Energy Efficiency Metric
The Green Grid has proposed a new metric called data center energy productivity (DCeP). Like PUE and DCiE, time is important to this measurement. DCeP can assess where work and energy are compared relative to a user-defined time limit.
Data Center Energy Efficiency=
(Useful Work Produced)
(Total Data Center Energy Consumption)
Useful work produced has been proposed as the (Sum of all tasks * Value of the task) * Time Based Utility Function * Absolute Time of Completion. Calculating DCeP allows users to right-size virtual and physical infrastructures to support business needs.
However, it is not an easy metric to implement especially for co-hosted data centers. Measuring the productivity from each lease owner would be tedious and impossible in certain cases. DCeP does not again include factors that take into account various energy saving strategies being implemented and will clearly not be sufficient.
Actions to be taken to improve energy efficiency at data centers may be classified into the following five items:
• Energy saving by efficient equipment operation — Reduce physical equipment and increase the work rate by consolidation, virtualization, etc.
• Energy saving by installing efficient IT equipment — Install IT equipment with even higher energy-saving performance.
• Energy saving for facilities — Reduce energy consumption for facilities by more efficient use of air conditioning equipment, more efficient use of energy conversion equipment, and use of devices that utilize the natural environment.
• Renewable energy penetration — Integrate green energy generated by data center efforts, such as photovoltaic power generation, wind power generation, and hydroelectric generation.
• Waste heat recovery — Incorporates waste heat recovery techniques from computer room air conditioning (CRAC) exhausts as well as chiller heat exchangers and minimizes the use of cooling towers.
• Sustainability / reliability — Introduce measure of real-time reliability and data center uptime.
• Personnel safety — Determine real-time safety hazards and include safety related events.
The new data center performance (DCP) metric is expressed by the function of the following five energy saving metrics and one of the advantages of this methodology is that each of the sub metrics can also be used independently.
PUE is the metric proposed by The Green Grid and widely used today.
PUE= (Total Facility Power)
(IT Equipment Power)
IT equipment utilization (ITEU) represents the degree of energy saving by virtual techniques and operational techniques using the potential IT equipment capacity without waste. Reduction of equipment to be installed is promoted by using the number of devices to meet the required IT capacity without waste. ITEU is essentially the average utilization factor of all IT equipment included in a data center.
ITEU= (Total measured power of IT equipment)
(Total rated power of IT equipment)
IT equipment energy efficiency (ITEE) is included in order to promote energy saving by encouraging the installation of equipment with high processing capacity per unit electric power. It is similar in concept to DCeP of The Green Grid. However, it is generally difficult to make actual measurements because a single data center contains a mix of various kinds of equipment and services. Therefore, a simpler approach is calculating ITEE using the spec values of the data sheet. For this reason, ITEE will be the weighted average efficiency of the energy saving performance in the catalogs of IT equipment in a data center.
ITEE= (Total Server Capacity+Total Storage
Capacity+Total Network Equipment Capacity)
(Rated power of IT equipment (Server+Storage+Network))
Green energy coefficient (GEC) has been introduced to promote the use of green energy, and from a power consumption reduction point of view it is positioned differently than the other metrics defined above. For every MWh of renewable energy produced means one MWh is not being produced by another generator and has a potential of reducing two-thirds of a ton of CO2 production.
GEC= (Green Energy MWh Produced)
(Total MWh Consumption)
Heat recovery coefficient (HRC) has also been introduced to promote the use of waste heat recovery technologies from a recycle/recovery point of view. It has been positioned differently than the other metrics defined above.
Waste heat of low temperature range (0°C to 120°C) could be used for the production of bio-fuel by growing algae farms or could be used in greenhouses or even used in eco-industrial parks. An eco-industrial park (EIP) is an industrial park in which businesses cooperate with each other and with the local community in an attempt to reduce waste and pollution, efficiently share resources (such as information, materials, water, energy, infrastructure, and natural resources), and help achieve sustainable development with the intention of increasing economic gains and improving environmental quality.
An EIP may also be planned, designed, and built in such a way that it makes it easier for businesses to cooperate, and that results in a more financially sound, environmentally friendly project for the developer. Temperatures in most data center hot aisles range from 80ºF to 115ºF (27ºC to 46ºC), still fairly low temperatures for some heat recovery strategies. Waste heat of medium (120°C to 650°C) and high (>650°C) temperature could be used for the generation of electricity via different capturing processes.
Some data centers utilize gas-powered micro-turbines to generate onsite power. During the winter, the 585ºF (307ºC) exhaust from the micro-turbines can flow through heat exchangers to produce hot water, which is then piped to a nearby office building to be reused in the building’s heating system. On an average, recovering the waste heat is estimated to save $235,000 per year for a mid to large scale data center.
HRC= (Waste Heat Recovered)
(Total Waste Heat Generated)
Equipment reliability coefficient (ERC) is included to promote the real-time tracking and calculations of equipment reliability combined with data center uptime.
ERC= (Calculated SAIDI )
System average interruption duration index (SAIDI) is calculated in units of hour per customer year and is an important index especially for cohosted data centers. Actual downtime would also be calculated in hours per customer year as well and based on actual downtime of the customer servers due to energy not delivered to these servers.
Personnel safety coefficient (PSC) is included to promote high safety standards being adopted in the data for personnel protection related to arc flash hazards.
PSC= (# of Buses>20 Cal/cm2 )+ 1
(Total # of Buses) (# of OSHA violations )
(# of Arc Flash Accidents)
Putting it all together
The beauty of the DCP metric is that it harmonizes metrics from around the world and incorporates them into a master metric such that only incremental activities need to be carried out to build upon existing PUE or DPPE indices used worldwide.
DCP is therefore =
(IT Equipment Utilization*IT Equipment Ability*Equipment Reliability*Personnel Safety)/
(Total Power Consumed-Green Power Used-Waste Heat Recovered)
DCP= ITEU * ITEE * 1 * 1 * 1 * DCRC * PSC
(PUE) (1-GEC) (1-GEC)
Real-Time Power Management
DCP defined above may not be approved by the industry anytime soon, however, intelligent power management software cannot only calculate each of these metrics in real-time but can also provide dynamic dashboards that may be published on websites worldwide so that managers can highlight data center optimization and energy savings techniques.
Power management software enables designers and engineers to conceptualize the power distribution model for mission critical facilities, simulate/test the integrity and security of the system, and analyze the results with accurate reports.
Putting a modern spin on the old adage "measure twice, cut once," IT and power engineers need to measure continuously to find energy savings. Organizations that are committed to conserving electricity and driving down operating expenses will have to focus on tracking DCP over time, managing the server and facility equipment load, and reporting actual results. Knowing how and what to measure is only the beginning. Knowing how to report and make accurate comparisons leads to the path of optimization.
A power management enterprise solution can be used to optimize power usage and energy savings while maintaining high availability and reliability. Optimize power usage to reduce the TCO and ensure a profitable life cycle for the mission critical facility.
Extending the power monitoring system by equipping it with an appropriate electrical system context, simulation modules, and playback routines will provide IT personnel with a powerful new set of tools. Finally, all of these capabilities should be included in one application with the flexibility and compatibility that allows you to expand and upgrade your power management system as your needs grow.