The goal of energy management is to make more intelligent use of energy in all its forms. CIOs and CTOs are concerned about energy management because data centers are a major consumer of power for running servers and other equipment and for cooling.
Energy management is a hot topic due to rising electricity prices rising and growing concern about sustainability. Today's practice of energy management often a shared responsibility. The VP in charge of facilities has one large part of the energy bill under his domain, and the CIO has the rest. Both roles have to work together in order to do the best job in energy management.
Of course, energy is just one of many different resources that must be managed efficiently. Carbon emissions are also under increasing scrutiny and water is likely to be the next focus. It is likely that the practices developed for energy management will be generalized into a comprehensive approach to resource management, a practice in which the use of everything consumed by a company, whether it's electric power, water, land, space, or transportation, will be more intelligently analyzed, optimized, and managed.
This problem statement is primarily concerned with how the CIO role can chart a path to a better state of energy management maturity.
Our goal for this problem statement is to examine how companies can start to do a better job of energy management; explore the stages companies typically negotiate as they improve their energy management capabilities and the ability to optimize energy use; and investigate how a program of incremental improvement could lead to a much higher state of energy management maturity.
Because this problem statement is focused on the CIO perspective, we're going to focus heavily on the data center, but we'll also cover things such as desktops and other IT-related energy consumption. The goal will be to develop a formal energy management maturity model to provide a method to understand the energy management challenges at a company and then address them in a way that provides continuous improvement and allows for fine tuning and the ability to address new challenges.
Special attention will be paid to the partnership between the CIO the VP of facilities. Both roles most certainly have to figure out how to work better with utility companies, which will express their desires to reduce power consumption through demand response programs, or by offering incentives for more efficient energy use, or for shifting use of energy.
Both will also likely have to interact with regulators, who will ask for more and more data about how energy management is used.
Context and Background
If energy management is not on the radar of a CITO today, it should be. It will be much less costly and painful to get ahead of this issue, rather than wait until you are asked about it and forced into a program of improvement.
We've done a variety of work with many different energy- and resource-management vendors, including SAP, Power Assure, nLyte, 1E, Intel Data Center Manager, CiRBA and JouleX, and we've found that all of them have a unique perspective, but all agree that there are some stages that most companies go through as they increase their energy management maturity.
The first stage of most maturity models involves increasing visibility. You can't really manage your energy or anything else well unless you understand current usage in all relevant dimensions. Visibility involves being able to monitor energy usage in the aggregate, something that surprisingly very few data centers can do, often because the power bill goes to the facility manager, not to the CIO.
Even when you can start getting an idea of your total energy usage from all the different sources, the next question becomes, which parts of your infrastructure are using what?
So the next step in visibility is the transition from understating your total energy usage to being able to allocate energy usage very specifically across your infrastructure in a variety of dimensions. Creating a detailed model of energy usage brings up interesting and unfamiliar problems.
For example, the stated or plated power consumption of certain equipment may not be close to its actual power consumption. The stated wattage limits of various equipment may not be close to their actual limits. You may be able to construct more accurate models of your actual, true capacity by testing and determining the actual power limits of the equipment, such as server racks.
In addition, in a virtualized world, you want to be able to know the impact of running a certain application on any servers for which you might consider reducing the power. Often companies have a difficult time understanding which applications are running on which servers, so that they can make an intelligent decision about when to reduce power, especially in the world of virtualization and cloud computing.
Analyzing Power Moves
Once you have visibility through a highly granular model, the next task is analysis. You must use that visibility to perform an analysis to understand both where energy is being used and what moves could improve efficiency.
This is a more complicated issue. You have to map the servers that are using power. You have to understand which applications are critical and which applications aren't. You have to understand which applications need more support at any given moment and which need less.
Then, you need to be able to reduce the amount of power to certain applications that may be in a low-impact state. The tactics for doing this are varied and quite interesting.
Intel Data Center Manager allows you to monitor very closely and control the power state of a chip. All of the vendors we mentioned above have different very interesting approaches to implement tactics, whether it's JouleX using network security monitoring techniques to understand what's happening on on a system, to Power Assure, which tracks power levels on myriad IT devices and provides a capability for dynamically adjusting a facility's power capacity to suit computing demand. The next level of maturity consists of the ability to make intelligent decisions tactically about energy management.
From this level, the next level of maturity is the ability to automate these decisions with minimal human input. The data center becomes a living thing, moving from one state of energy usage to support high levels of performance to lower states in which equipment is quiesed and put to sleep.
The nirvana of energy management is the ability to design your IT infrastructure with energy management in mind, so that all of this visibility, automation and optimization takes place as part of the way you do business.