Whether you are a large company or small technological cooperative, the role of the data center may be central to the functioning of your business. Unfortunately, powering racks and racks of servers and cooling them can be both expensive as well as environmentally hazardous due to the high power demands of most information communication technology (ICT) industry. Revenue in the global server market grew 3.1% during the first quarter of this year, and is projected to surpass 2011 revenue levels by 7.1% over the course of 2012.1 While there is no current plan to eliminate the footprint of data centers,2 there are steps that companies can take to lower their carbon footprint through organizational and managerial approaches on the higher and lower levels.
3
BENEFITS FOR THE ENVIRONMENT: The typical emissions profile for a large corporation such as Facebook can amount to 628.3 million pounds of carbon dioxide. Furthermore, the current 80 metric ton carbon dioxide emissions level from data centers are due to go up to 340 metric tons by the year 2020.4 Data center power usage of Facebook alone was 532 million kilowatt hours, with Google taking a grand prize of 2.26 million megawatt hours from its data centers alone.5 By implementing steps to prevent outrageous power loads and perhaps by including a greater proportion of effects of climate change as well as establish a precedent for sustainable data center management.
BENEFITS FOR YOUR WALLET: Using less energy for your data centers will inevitably save you money on electricity bills. In the year 2000, data centers used an estimated 12 billion kilowatt hours of electrical power, which then increased to 23 billion in the year 2005.6 Today, growth in the industry has increased these figures significantly, and further highlights the need for power savings and associated cost cutting.
Time and Effort: Medium – High. Establishing green data center policy requires cooperation from many levels of management as well as training for employees. Not only is there a need to set proper power reduction targets and establish good monitoring policy, but a need for employees to learn and gain incentive from a greener data center culture.
Cost: Although there may be an initial cost for more efficient equipment and restructuring, there is a much greater overall benefit. This should save you money over the long term!
Instructions:
- Evaluate your energy efficiency and energy portfolio. The first steps towards finding a solution to the problem is to actually be able to perceive the problem, and a quick look at your corporate energy bills from data centers may make a convincing case in finding ways to reduce this cost as a company. Try to plan some sort of road map for cost reduction by setting appropriate targets. This is of course not an easy task, and varies based on the function and size of your corporation. Additionally, try to promote initiatives that boost alternative energies on your energy portfolio. Many state sponsored utility corporations offer energy rebates for power savings and greater usage of alternative energies. PG&E in California, for example, offers conditional rebates in conjunction with the California Public Utilities Commission given certain energy standards are met in respective data centers.7
- 9Reconsider Redundancies in server systems and cooling modules. Consolidate your servers since about 10 – 30% of servers can be turned off, which could amount to a cost savings of an estimated $1200 per server per year and another $300 – 600 in cooling costs. PG&E actually offers a virtualization incentives program that pays $150 – 300 per server removed.8 Many times, redundant cooling systems are put into use to account for peaks in performance levels. Using such measures usually results in maximal outputs when only a fraction is needed. Using modular or scalable equipment usually provides some relief in remedying this inefficiency issue. Scalable blade servers can concentrate more computing power in a smaller space, requiring less power to cool. Sometimes simple maneuvers can result in major energy savings. Try to enclose servers in designated hot isles such that cooling can be concentrated in these isles. Reduce air flow loss by sealing off holes in your walls, floors and ceilings. Make sure there are plenty of air ducts near your cooling apparatus.10
- Monitor and turn on power management tools on servers and terminals when applicable. Encourage monitoring for down time activity, and not just uptime and high performance. Most windows and linux boxes have a power management system and a power saver feature that can reduce power requirements at least for terminals. Try to understand ACPI compatibility with your systems’ operation, and try to implement its power management standard if possible11. ACPI – Advanced Configuration and Power Interface – is an open hardware standard that checks, consolidates, and improves power management for hardware devices.12 Using efficient power supply units is also crucial for proper power management practices. Inefficient units shipped by default with many servers tend to waste a lot of energy. Most server manufacturers will sell servers with power supplies that overshoot the requirement (600W for a 300W need), these types of practices are being phased out nowadays, but it is an important detail to be aware of. Although more efficient power units cost 15 – 20% more, they could save up to 80% when implemented and save a good deal in the long run. In fact, 80% efficiency should be the benchmark for buying better power units as every $20 spent saves $100 in the long run at this level of efficiency.13 Google has taken better power design to a new level with 12-volt standardization across the board, while maintaining higher data center operation temperatures with improved peak performance.14 Although a re-design of a server set-up can amount to more efficient operation, this move would be costly and depends upon the needs and size of your respective company.
- Be aware of new technologies and certifications that can be utilized in data centers. Chip makers such as AMD are developing processors that can offer 65% in energy saving for 50% CPU utilization.15 Furthermore, the availability of more efficient servers such as Intel’s Xeon family of servers has demonstrated a 40% decrease in power consumption. New server technologies try to consolidate I/O and memory controlling onto the same chip, this saves a lot of power on the long run. There is also the option of virtualization for smaller server systems that removes the physical existence of the server and replaces it with a virtual entity that runs on a terminal window. The idea is that small server systems can be consolidated into larger ones for operation when server tasks are well monitored and distributed. The virtualization device will keep track of your server space and remap applications to different physical locations as necessary to achieve optimal efficiency.16
17
The Energy Star ratings system is an index of electrical devices that follow certain standardized criteria for low carbon emissions. Although servers are not exclusively energy star certified, a lot of server components and cooling systems are energy star rated. Purchasing energy star rated components can save costs on the long term, while reducing your corporation’s carbon footprint.18 - Try to minimize internal barriers while establishing good energy management policy. Although much more simply stated than done, fostering open communication with facilities and IT working groups can help institute good energy saving policy.