Green IT: The Gift that Keeps on Giving
We talk with IBM Senior Fellow John Lamb about green IT's many benefits.
Companies with deep pockets are pouring vast resources into delivering more efficient ways of powering data centers. Among them are i/o Data Centers LLC, Bloom Energy and IBM Corp. (For more on these companies' efforts, see "More Data, Less Juice,") In late February, IBM launched its Smarter Buildings effort. IBM Senior Fellow John Lamb is one of the world's foremost authorities on green computing. The author of "The Greening of IT" (Pearson, April 2009) -- inspired by on the famous counter-culture book "The Greening of America" (Random House, 1970) by Charles A. Reich -- Lamb's book is a soup-to-nuts treatise on creating an energy-efficient data center. Redmond Editor in Chief Doug Barney asked Lamb how to save money on power while keeping the data center humming.
Q Energy efficient technologies have the promise of a payback, but many IT budgets have shrunk. How can IT move toward efficiency while living within these budget constraints?
A Green IT is perhaps the best way for any organization to get started with energy efficiency. IT budgets will always provide for refreshing servers, laptops and other IT equipment every three or four years as part of an organization's strategy to keep up-to-date with technology. This refresh policy provides a great opportunity for a company to buy new, energy-efficient equipment and implement server and data storage virtualization. Virtualization provides significant IT flexibility, reduces data center space requirements and can reduce system management costs. The savings in energy costs becomes significant -- up to 50 percent -- on top of an already robust business case for virtualization, and this is without even considering the reduction in electricity costs.
Q For a typical data center, is there one technology that offers the most efficiency benefit?
A The most significant step most organizations can make in moving to green data centers is to implement virtualization of servers, data storage, and clients or desktops. There's also a virtual IT world of the future -- via private cloud computing -- for most of our data centers. Although the use of cloud computing in your company's data center for mainstream computing may be off in the future, some steps toward private cloud computing for mainstream computing within your company are currently available. Server clusters are here now and being used in many company data centers. Although cost reduction usually drives the path to virtualization, often the most important reason to use virtualization is IT flexibility. The cost and energy savings due to consolidating hardware and software are very significant benefits and nicely complement the flexibility benefits. The use of virtualization technologies is usually the first and most important step we can take in creating energy efficient and green data centers.
Q How do you prioritize green investments?
A IBM's Enterprise Computing Model for reducing thousands to about 30 large centralized servers provides a process for prioritizing consolidation steps. Those prioritized steps are: Migrate servers delivering largest savings first (for example, stranded infrastructure). This primes the pump and generates enthusiasm and savings for other Green projects; Eliminate assets with lowest utilization first. These assets are not pulling their weight when measured by watts/logical image or other common metrics to compare servers; Identify assets with an upcoming compelling event to mitigate expense (upgrade, move, asset refresh). It is always easier to have a positive ROI and be Green within the normal refresh of assets; and aggregate by customer work portfolio to leverage strong customer buy-in. Ease of migration assists speed and successful workload migrations. Start with oldest technology first since it uses the most power and provides the least performance. Focus on freeing up contiguous raised floor space. This enables growth and the addition of energy efficient new IT and facilities equipment.
Q At this point, what percentage of apps can be virtualized? Which ones would be the last to make that move?
A Almost all applications can be virtualized. However, depending on the virtualization engine, there are certain applications that can cause problems. We had performance problems with a data warehouse that was on VMware. If the virtualization engine is a big Unix box with hardware designed to give each virtualized server all the I/O bandwidth it needs, then a data warehouse on a virtualized server might not be a problem.
Client, or desktop, virtualization offers a great potential in energy savings. However, the benefits versus the risks can vary greatly from company to company. Various studies have estimated energy savings of more than 60 percent by using client virtualization. Client virtualization -- often called thin-client computing -- is not a new concept and goes back at least 15 years. In fact, thin-client computing, where the server does all of the computing, is similar in concept to the terminals we used to connect to the mainframe before the advent of the PC. Thin-client technology can be a significant benefit, for example, in supporting help desks where everyone at the help desk needs to access the same server applications. But the major risk moving to thin clients is loss of flexibility.
On some operating systems, such as Windows, software products are designed for personal computers that have their own local resources. Trying to run this software in a thin-client environment can be difficult. So client virtualization through thin-client computing gives us very significant benefits, but there are also concerns. A good place to start with client virtualization is the help desk, where the benefits usually greatly outweigh the concerns.
Q Are there good calculation methods, return on investment (ROI) models or general approaches that help IT choose which projects to invest in?
A In consolidating servers to save money and energy, more than a single methodology needs to be applied to get the fewest number of servers. IBM's Enterprise Computing Model for reducing thousands of servers include several steps related to ROI. The biggest savings of all for going green comes from creating a culture and infrastructure that exploits technology for creating a sustainable data center. Conducting or having a third party conduct an energy audits can benchmark where you are and identify projects with ROIs that can be prioritized to give cascading green returns.
For every watt you save in IT equipment you reduce the infrastructure UPS [Uninterrupted Power Supply] and cooling loads and generate savings for future projects. Conserving energy in the data center allows the dollar savings to be used for adding more value to the business. Use energy like a precious commodity. Turn it up or on when needed and throttle back or turn off when not needed. Create a culture and data center that's intrinsically green.
Newer generations of servers and storage are built with more efficient power supplies, processors, memory and I/O. Just about everything in newer servers and storage provides more performance or stores more data with fewer watts. Decommissioning older servers that never were designed for virtualization or energy efficiency can be one of the most cost effective ways to green your data center.
Q How does one become a green IT expert, and does this translate into IT leadership and career movement?
A Many energy standards and metrics are still in the development stage. With all the research going into energy efficiency, we should expect continuous improvement not only in technology, but in the government and industry standards and metrics used in determining the goals for green data centers and green IT in general. All of the activity going into green IT translates to significant opportunities for a long time to come in IT leadership and careers in green IT.
Q How does one engage upper management in these decisions about data center and server room efficiencies?
A For data centers, the cost saving incentives are so great that it's usually easy to engage upper management in data center energy efficiency. The environmentally friendly aspect of green data centers is an added plus. While there might be some 'green washing' by exaggerating a company's dedication to data center efficiency, companies are motivated by the real economic benefits they've seen. A typical U.S. data center with 25,000 square feet uses approximately $2.6 million in energy costs per year. Improvements in energy management can save up to 50 percent of those costs. A million dollars in savings is a motivator to drive sufficient interest.
Q Is cloud computing mature enough to be a viable route to green computing? Which apps are best suited for the cloud today?
A Cloud computing is an alternative to having local servers or personal devices handling users' applications. Although the early publicity on cloud computing was for public offerings over the public Internet by companies such as Amazon and Google, private cloud computing is starting to come of age. A private cloud is a smaller, cloudlike IT system within a corporate firewall that offers shared services to a closed internal network. Shared services on the infrastructure side (such as computing power or data storage services) or on the application side (such as a single customer information application shared across the organization) are suitable candidates for such an approach. Of course, IT virtualization would be the basis of the infrastructure design for the shared services, and this will help drive energy efficiency for our green data centers of the future. One of the best environments for a private cloud is to manage test environments. For a current customer, there are challenges in efficiently managing test environments. There's not a good scheduling process for shared resources and there's a significant amount of test server waste because of lack of governance and management. A private cloud has been proposed to manage the company's virtualized test environments. The private cloud will allow tight governance around automated provisioning and de-commissioning of environments allocated to projects. IBM Cloudburst technology is being proposed.
Q How important is the notion of simplicity, and how do you simplify all elements of your data center?
A Simplicity will help give us the fewest number of servers. IBM's Enterprise Computing Model for reducing thousands of servers to about 30 large centralized servers used basically a simplification approach to consolidation. Going back to the basics and using tried and true simple approaches to saving energy always works well. A good example is the use of 'free cooling.' Outside air can substantially reduce energy required by computer room air conditioners. Data center site selection will enable more days of free cooling when the climate provides a big difference between day and night temperatures. Colorado is an example of an excellent climate to exploit free cooling.
Q How does one design green technology into future IT planning and purchases?
A The economic benefit of green IT provides the motivation to design for energy-efficient IT planning and purchases. This motivation is worldwide. As I'm now based in Johannesburg, South Africa, I can tell you the daily papers have articles on how the electricity costs for South Africa are scheduled to climb over 30 percent a year for the next three years. Using alternative technology for energy generation -- such as wind, hydro, landfill-gas and solar -- is becoming very attractive. Also, the company I'm working with in Johannesburg has been told by the local power company that they can't add any more servers to their data center since the power currently being supplied is at a maximum. Green IT is important worldwide on several fronts, including energy cost, environmental aspects and company limits on power available to data centers.
Q How does one convince all stakeholders that data center efficiency is worth committing to?
A All companies are dependent on IT, so the incentive to make their IT operations more efficient provides an opportunity to also reduce energy use. Improving IT efficiency will always reduce IT costs and energy use. However, improving the efficiency of your IT systems in order to reduce costs provides a great opportunity to baseline and provides ongoing energy measurement to determine and trend energy use improvements.
Q How do you decide whether to upgrade an older data center or build a new one using the latest green techniques?
A Energy-saving technologies, such as server and data storage virtualization, can produce a 50 percent energy savings for new equipment in any data center. Often basic data center equipment, such as the UPS used for backup in case of a power outage, is replaced after many years as part of a company's plan to improve data center reliability. As power for all IT equipment flows through the UPS, replacing an 80 percent efficient, 20-year-old UPS with a new 95 percent efficient model will provide significant energy savings -- even though the motivation for replacement was based on improving UPS reliability. There's usually no need to build a brand-new data center in order to gain significant energy efficiency improvements.
Q Cabling sounds like no big deal, but it actually is. Can you explain why?
A The explosion in the number of servers that data centers must support has created cable-management challenges in many facilities. If not properly managed, cables can obstruct air flow through perforated floor tiles and prevent air from being exhausted out the rear of the rack. Check the under-floor plenum to determine if cabling or piping is obstructing air flow. Overhead cabling is becoming increasingly popular; it eliminates the potential for obstruction. Deeper racks are now available to allow for increased airflow. Sometimes existing racks can be equipped with expansion channels to add depth for cables and airflow. Be cautious when using cable management 'swing arms,' as they're not compatible with all IT equipment air flow patterns.
Q What are some of the smartest things you can do to make your cooling systems more efficient?
A Data center cooling specialists such as Emerson-Liebert have a great deal of experience that should be leveraged. Here are some of those proven strategies on increasing data-center cooling efficiency:
- Proper sealing of the data center environment. A vapor seal plays a critical role in controlling relative humidity, reducing unnecessary humidification and dehumidification.
- Optimizing air flow. Rack arrangement, computer room air conditioner placement and cable management all impact the amount of energy expended to move air within the critical facility.
- Using economizers where appropriate. Economizers give us free cooling by allowing outside air to be used to support data center cooling during colder months, creating opportunities for energy-free cooling.
- Increasing cooling system efficiency. New technologies, such as variable capacity systems and improved controls, are driving increased efficiency of room air conditioning systems.
- Bringing cooling closer to the source of heat. Supplemental cooling systems bring cooling closer to the source of heat, reducing the amount of energy required for air movement.
Together, these methods can reduce cooling system energy costs by 30 percent to 45 percent and generate significant, recurring savings. Coupled with emerging technologies such as higher-efficiency processors and new chip-based cooling technologies, these measures can keep energy costs in line as server densities and the price of energy continue to rise. A major opportunity for many data centers is to work with the building management system and use waste heat from the data center in the winter to heat offices. The data center cooling system would act as a big heat pump.
Q How do you apply the concept of measuring to greening your data center?
A One of the most important arenas for development of green IT is that we must have better measurements in order to better manage energy use at data centers and throughout the corporation. We'll see great strides here, like gas mileage monitoring on the Honda Prius hybrid. Currently we can use data network 'sniffer' concepts to determine network bandwidth usage and response time for each user of a server. That same sniffer concept will be developed for power and energy used. There will be inexpensive power- and energy-monitoring appliances that will act like the network sniffer and give us information on the power and energy use for each IT device. That appliance will send the information to an energy-monitoring system (a server). Then we can see the actual energy-reduction results of using virtualization and other energy-saving initiatives. The electric power and energy monitoring device will also be in our homes and will have the capability of sending information to our laptops. Then we'll be able to see the actual power and energy-use history of our refrigerators or window air conditioners. We then will be able to better manage our energy use by measurements and trending.
Q What areas of innovation is IBM responsible for?
A IBM has been very innovative in the automation of measuring and billing of energy consumption. This makes energy usage part of cost and green decisions. Without energy and cooling knowledge, requirements are unknown, inaccurate and often over-planned, leading to inefficiencies. An example of new technology to optimize energy use is IBM's Active Energy Manager. Monitoring energy usage and developing trends is key to understanding how energy is being used. This first step to optimizing energy use opens up the potential to become more efficient and optimize for performance and watts.