The problem in power and cooling, aside from the continuing increases in costs, is generally not having enough - power coming into the datacenter, circuits on the UPS or battery backup, and/or cooling ability in the data center and of the air conditioning system (computer room air conditioner or CRAC). In addition, remember that cooling requires additional power to run the cooling system, so it's a double whammy.
Consolidating servers through virtualization allows you to reduce the number of physical servers in the data center, sometimes significantly, and therefore reduce all the power and cooling requirements, as well as physical space. Consolidating onto blades further reduces the amount of power and cooling required per server (over a comparable rack mounted server) and offers even better space utilization. However, the increased density of blades will require more cooling per footprint for the blade portion of the data center. In fact at a rack level, blades can represent an increase from 2-5kW per rack of legacy servers to 20+ kW for a rack of blades. Overall, consolidation with virtualization and blades can offer significant help with both space and power and cooling, but awareness of power/cooling requirements and careful planning are essential.
Power and cooling reductions and increases
When estimating power consumption after consolidation, it is important to estimate both the number of physical servers, and their power rating, since the new servers probably will be more powerful and will have a higher power rating.
The numbers on the server nameplate indicate maximum power consumption for that server. Since power consumption is related to utilization, this number will vary with workload and utilization, up to the maximum.
Since an idle server still uses power, power consumption does not correlate directly to utilization. APC recommends using a steady state of 67% as an average.
Check with your power company on possible incentive or rebate programs around virtualization and consolidation. For example, last November, Pacific Gas and Electric launched a program offering users incentive payments of $150 to $300 per server eliminated through consolidation. Users are required to conduct an inventory before they start their consolidation project and apply to PG&E for the incentive program. Eventually this program will probably become a rebate program, where users could simply send their reports after the fact and receive a rebate (the amount per server of the rebate has not yet been determined). According to VMware, they are working with other utility companies around the country on similar offerings.
There are a variety of tools and services available for planning your consolidation project. Check with your server and virtualization vendors, power and cooling partners, and reseller/integrator for tools such as power calculators and capacity planning tools. VMware, for example has a capacity planning tool which can be used by their professional services group to take an inventory of your current environment and monitor it over a period of time, including hardware and software utilization, peaks, etc. They can then use the information to recommend phases of implementation for consolidation and for new equipment recommendations.
Look for help from your partners in planning for power and cooling needs, both the reductions and increases based on fewer servers running higher powered processors, with a higher density form factor. Take advantage of power and cooling audit and assessment services and get recommendations on how to make your data center energy efficient as you implement these new technologies going forward.
A number of vendors are now working together to help address power and cooling issues. AMD, APC, Dell, HP, IBM, Rackable Systems, SprayCool, Sun Microsystems, and VMware joined together as founding members in an organization called the Green Grid, an association of information technology professionals seeking to lower the overall consumption of power in datacenters around the world. Their goals include developing and offering best practices on efficient power management in data centers, promoting energy efficiency initiatives, and developing standards and methodologies for measuring power.
Blade vendors have made great progress in the current generations of blade systems, improving power and cooling efficiencies significantly. IBM says that they have increased the efficiency of their power supplies from 65% efficiency in their 1U servers to 91% efficiency in BladeCenter H. (65% efficiency means that they convert 65 % of the power at the wall and the rest goes into room in the form of heat. In other words, for each 1kW from the power provider, 650 W goes to the server, and 350 W goes into room). Blade vendors are continuing to work on heat issues, with a variety of options for cooling at the blade, rack and room level. Liquid cooling continues to be a controversial option for many, though chilled water has been a part of IBM mainframe data centers for decades.
In addition to the improvements in power and cooling efficiencies, there are enhancements now and coming soon in software management capabilities. IBM and HP, for example have adding new levels of intelligence in their blade server management capabilities to monitor thermal status, and actively manage components when thermal problems occur. VMware is planning additional intelligence in their Dynamic Resource Scheduler (DRS) product, allowing users to automate processes such as dynamically consolidating workloads to fewer servers, and powering down inactive servers, based on DRS user-configured rules. In addition, rules might also instruct DRS to power down memory or reduce CPU frequency based on utilization and inactivity. These types of features are available today in certain other hardware offerings, such as HP's Integrity products, but not yet in their blade offerings.
Clearly, power and cooling issues are hot priorities for data centers today, with users looking for help from hardware, software, and power companies, as well as resellers and integrators. Virtualization and blades can both directly contribute to addressing these issues. There are still many misconceptions about whether blades help or hurt in this area, often based on old data. Choose your partners carefully, and make sure you're working with up-to-date information. If you do, and you plan well, consolidating through virtualization on blades can save you big bucks in the short and long term.
For more information on implementation, tips, planning advice, ROI analysis and more on these issues, check out the Blades and Virtualization Summit on May 1-3. Hear from blade and virtualization vendors and consultants, power and cooling specialists, analysts, TechTarget expert panelists, and other experts on how to address these issues in your environment.
About the author: Barb Goldworm is president of Focus Consulting, a research, analyst and consulting firm focused on systems and storage. She has spent thirty years in technical, marketing, industry analyst, and senior management positions with IBM, Novell, StorageTek, Enterprise Management Associates, and multiple successful startups. She chairs the Server Blade Summit conference on blades and virtualization, and has been one of the top three ranked analyst/knowledge expert speakers at Storage Networking World. Barb has authored numerous business and technical white papers and articles and has just finished a book, Blades Servers and Virtualization: Transforming Enterprise Computing While Cutting Costs. Barb can be reached at email@example.com.