The unit step function makes the Loss Function a one-sided loss function, so-as not to penalize performance exceeding the target value. More comprehensive and sophisticated metrics could be derived, however there is a move internationally to regulate carbon emissions and active trading markets are developing, putting a real value or cost on these emissions.
Equation 4 requires conversion of electrical power consumption to carbon emissions. While this conversion also varies by region and over time due to the mix of generation sources, this paper uses a conversion based upon the United States national average ENERGY, The U.
The desired PUE number by is 1. Equation 3 assumes that there is an economic cost or value that can be assigned to carbon emissions based on a market in Carbon Credits. Converting CO2 emissions to currency values using carbon credit equivalents are subject to location, market pricing and energy usage. Table 2 illustrates the current variation in carbon credit prices.
The large degree of variation observed in the prices is reflective of lack of standardization in the basic definition of the credit, as well as lack of regulation in the markets. As these markets become more regulated, there will still be variation in value over time, as well as between different economies as they try to meet their international commitments to reduce emissions.
However, within a particular economy, at a point in time, the variation in pricing should be significantly less. The fluctuation in outside air temperature influences cooling demands, which effects energy consumption and introduces variation in the data center PUE. Conversely seasonal e. This predicates a need to do real-time data collection and PUE calculations. The situation of less than ideal power metering locations available at the RIT data center is likely common with older data centers.
Figure 1. This figure shows predictable seasonal fluctuations in current draw, with peak consumption during summer months and high demand periods around the end of the school year.. The yearly average of power consumption at point 1 is approx. The observed seasonal variation results in a shift in consumption of about 60A or RIT has an inexpensive and unique data center temperature monitoring system using single wire sensor technology and a php web site that provides real-time temperature data.
This data can be used as to continuously monitor performance of the HVAC controls, and also look for hot spots that represents high server consumption or ineffective cooling layout. Heat map of the RIT data center Operating the data center at a higher mean ambient temperature offers opportunities to reduce the energy consumption associated with cooling. Figure 3 below provides an overlay of the data center temperature with the data center current draw, illustrating the variation in these parameters over a one- day period in February of Figure 3.
Power Statistics from Point 4 in Real Time 3. The data collection systems described above were used for the engineering study. The associated reports provide UPS and generator capacity, total annual power usage, and also break out the HVAC associated power and other auxiliaries.
The data summaries from the report were used for the data center analyses that follow. Table 3 below shows an analysis of RIT data center power consumption for and , broken out by sub-system, and also total energy use.
Table 3. Table 4 show the carbon emission projection for the RIT data center. Table 4. This provides an additional economic penalty for excessive deviation from best practice. Table 5 also gives the calculated Taguchi loss for the RIT data center through , assuming no changes in the existing infrastructure. Table 5. Upgrading to higher efficiency IT equipment reduces total energy consumption but may actually result in an increase in PUE. Increasing the overall data center energy effectiveness reducing non IT related power results in PUE reductions and associated energy savings.
Below are a few examples of changes that can make a positive impact on PUE examples are relative to the baseline. Reduction in lamp fixtures from 19 to 6 results in an improvement of approx 0. If in addition to turning the reduction in operating fixtures to 6, the remaining fixtures are converted to LED lighting, PUE is reduced by approx 0.
Rochester New York is in a temperate climate and can utilize free air- cooling for approximately 5 months of the year.
The 45 deg passive cooling limit is set based upon the targeted data center interior control temperature of 65 deg F.
If the data center control temperature is raised to 75 deg F, which would allow passive cooling for approximately 7. This improvement results in a PUE improvement of 0. Increasing equipment operating temperatures raises a concern of reduced equipment reliability or durability. Experience at RIT indicates that the current physical life of IT equipment is longer than the useful life due to technology obsolescence.
The standard form of the Taguchi loss function penalizes adverse process behavior and variation relative to a performance benchmark. We have suggested that the CO2 emission associated with benchmark data center PUE is an appropriate benchmark. The employed form of the Taguchi loss function penalizes data center performance deviation that significantly exceeds the variation between best practice and state of the art benchmarks, resulting in increasing cost associated with non-compliance.
These results indicate that the RIT data center infrastructure is out of date and requires significant improvements, if not complete overhaul, in order to achieve best practice and zero the loss function.
The loss function formulation also results in no economic penalty when the data center is at benchmark performance. The justification for this is that advancement beyond best practice is often not cost effective and there may be better more cost effective opportunities to reduce carbon emissions elsewhere. Consortium Green Grid. Consultants, R. Deming, W. Di Mascio, R. The economic assessment of process control quality using a Taguchi-based method.
Journal of Process Control, 11 1 : Public Law, Feng, W. The origin and evolution of green destiny. Greenberg, S. Best practices for data centers: Lessons learned from benchmarking 22 data centers.
Montgomery, D. Statistical quality control: a modern introduction. Pinheiro, E. By using green technology, large IT corporations will take a step ahead in saving the environment and make the earth free from any danger.
Destroying environment any further can lead to situations which can be hazardous and irreversible. Green Technology helps reduce the use of fossil fuel. It is expected that energy production from green technology will be higher than fossil fuel sources of energy like oil, and gas in the future. Using green technology will become not only important but mandatory too in the coming years because of limited Non Renewable energy sources.
Green technology encourages the concept of cleanliness, freshness and also promotes new dimensions. The sooner we realize the importance of using green technology, the better it will be for our planet and its environment.
Shows methods of Green Computing In this age of ever advancing technologies, new efficiencies are developed within the equipment ensure that most of the energy goes into the IT load, instead of being lost in the powering and cooling of the equipment. Environmentally friendly technologies are coming into their own, for example, designing datacenters powered by renewable energy sources. As 21st century belongs to computers, and electronic items, energy issues will get a serious ring in the coming days, as the public debate on carbon emissions, global warming and climate change gets hotter.
Although this development has helped the human race, mismanagement has led to new problems of contamination and pollution. Faster processors historically use more power. Inefficient CPU's are a double hit because they both use too much power themselves and their waste heat increases air conditioning needs.
The technical process acquired during the last century has posed a new challenge in the management of wastes. Green Computing Green computing is the practice of using computing resources efficiently. Modern IT Systems rely upon a complicated mix of people, networks, and hardware, as such, a Green computing initiative must be systemic in nature, and address increasingly sophisticated problems. Green computing has become the utmost requirement to protect environment and save energy, along with maintaining operational expenses in today's increasingly competitive world.
Currently large IT corporations are working on implementation of the green computing practices. It has been seen that the power consumption of the green energy data center system is half that of a conventional data center, and green energy accounts for another half for driving.
The data center is green and energy saving, and the profit margin of the data center is improved. What is Green Data Center? A green data center is an enterprise class computing facility that is entirely built, managed and operated on green computing principles providing the same features and capabilities of a typical data center by using less energy and space. It can be considered as a transformation that encompasses technical innovation, operational improvements, new design principles, changes to the relationship between IT and business, and changes in the data center supply chain by using Green Technologies.
Basically a green data center comprises of a utility power source, a facility electrically connected to the utility power source, a renewable energy provider and an IT equipment electrically connected to the renewable energy provider, wherein the total power received by the IT equipment is not less to the utility power provided by the utility power source thereby achieving the effect of carbon-reducing and energy-saving.
This step includes monitoring and measuring the consumption of energy for the entire enterprise The e a e et i s that a help i a al zi g the g ee ess of a e isti g data e te. Based on this measurement, corrective measures can be taken to transition to a GDC.
Carbon Foot Print A carbon footprint is historically defined as "the total sets of greenhouse gas emissions caused by an organization, event, product or person. Carbon footprint spectrum of definitions ranges from direct CO2 emissions to full life-cycle greenhouse gas emissions and not even the units of measurement are clear.
By customizing innovative power and cooling strategies to the unique climate where our data centers operate. There is more emphasis on the power and cooling infrastructure. The power scheme being used currently in the enterprise will change with the increase and decrease in usage of energy. Apart from the benefit of reducing the operating costs by reducing energy costs, practices that reduce consumption of energy are also being used. Different Methods Of Greening a Data Center Green strategies and technologies exist today to help optimize space, power, cooling and resiliency while improving efficiency and reducing costs at the same time, helping to position companies for growth and enabling them to meet expanding business needs.
Although the e is lea l o si gle ight a to eate a g ee data e te , e pe ts elie e that the most productive first step is to conduct a best practices assessment and energy audit. Turning off servers that are not doing any work and also by turning off computer room air conditioning units in areas that are overprovisioned for cooling.
Most important aspect to be implemented is reducing cooling requirements of the data center. There are a number of factors that should be considered in developing a plan for improving power and cooling efficiency by reducing the heat generated in the data center. Improvements in rack and room layout can increase energy efficiency with relatively low upfront investment.
Water-side economizers Water-side economizers, utilize outside air to directly cool the chilled water and can further reduce the energy required to run the data center chillers.
Recycling of energy can be done by augmenting capacity and efficiency of chilled water systems with thermal storage systems that store energy generated at night, when chillers typically operate more efficiently, and then release this energy during the day.
When the systems are larger and more amenable to taking advantage of no-cost cooling, outside air temperatures are sufficiently low to provide some or all of the cooling requirements. Companies can save energy and gain cooling capacity by relaxing stringent relative humidity and temperature requirements for their data centers.
Refrigerants used for cooling Safety becomes a key concern when the source of cooling moves close to sensitive electrical equipment so refrigerant is an ideal choice for high-density applications. Because refrigerant turns into a gas when it reaches the air, so it will pose a safety hazard and a leak would not damage IT equipment. Refrigerant solutions also provide an incremental energy efficiency savings of between 25 percent and 35 percent, based on kilowatts of cooling capacity per kW of heat load.
Dynamic Monitoring and Management module is used for monitoring and managing the cooling and energy efficiency of data center. After taking a measurement survey, IBM combines sensors and measurement survey results for ongoing analysis and reporting.
It is estimated that up to 35 percent of an average air-cooled data center's energy consumption and carbon footprint is by powering the necessary cooling systems to keep the processors from overheating and not by computing itself. Water has been embraced by IBM as a coolant after understanding the limitations of air cooling.
With conductivity properties up to 3, times better than air, water offers a promising path toward sustainability. Traditionally, the heat exchangers have operated using chilled water. IBM researchers in the company's Poughkeepsie, N.
Department of Energy as part of a two-year research project, which is drawing to a close. The project sought to increase the efficiency of the world's 33 million computer servers and resulted in 21 IBM patents for technology known as heat-exchangers, which can cool computer servers using unchilled tap water, an innovation that can slash in half the amount of energy used in data centers. To keep the mainframe computer from overheating, IBM created a specialized water- cooling system.
There, IBM researchers introduced an intelligent water-cooling circuit under development for chips. When compared with air-cooled data centers, the intelligent water-cooling circuit not only reduces energy consumption by 40 percent, but also makes waste heat available for direct reuse, such as for heating homes.
IBM reported that the first prototype system was already reusing three- quarters of the energy needed to operate the data center. The system, dubbed Aquasar, reportedly consumed up to 40 percent less energy than a comparable air- cooled machine. Through the direct use of waste heat to provide warmth to university buildings, Aquasar's carbon footprint was reduced by up to 85 percent.
As a result, companies can save more energy and costs by not having to chill or heat their water for the data center. Examples Solar-Powered Data Centers, Wind-Powered Data Centers, Geothermal Data Centers and further other sources like a photovoltaic thermal hybrid solar collector, a bio-gas power generator system, a bio-oil power generator system electrical power source can be used.
Powering data centers with renewable energy sources like wind or solar is an obvious solution to global energy problems. Here are some companies who have announced plans to power their data centers with renewable energy. With Onsite Biogas and Fuel Cells, Microsoft Data Center Power plants operating on renewable biogas efficiently convert a waste product into ultra-clean electricity and heat, solving economic and sustainability challenges for environment.
Biogas is generated by the decay of organic material i. This decaying organic matter releases methane, or biogas. A harmful greenhouse gas, the biogas cannot be released directly into the atmosphere, while flaring creates pollutants and wastes this potential fuel source.
Capturing and using biogas as a fuel to generate power solves these challenges in a carbon-neutral manner. T pi all , atu al gas is the fuel that feeds the electrochemical process. Microsoft has opened a zero-carbon, biogas-powered data center in Wyoming that combines modular data centers with an innovative way to leverage waste from a nearby water treatment facility.
This data center was built alongside a wastewater treatment plant, where anaerobic digesters break down to produce the gas that feeds the fuel cells that keep the Data Centers Functioning. FuelCell Energy of Connecticut has developed the fuel cell technology to convert unused biogas into ultra-clean power generation solutions.
Siemens worked with Microsoft and FuelCell to engineer and install power monitoring equipment for the data center of Microsoft. Data center containers filled with servers are deployed next to a water treatment plant in Cheyenne, Wyoming. Servers are powered using electricity from a fuel cell running on methane biogas from the plant. The plant uses an electrochemical reaction to generate electricity and heat. Virtually no air pollutants are released because of the absence of combustion.
Each fuel cell generates around kW of renewable power of which the data center uses about kW.
0コメント