Written by: Michael Harding Sr., Product Marketing Manager
The world is recognizing that data centers are big users of energy, and many corporations are now dealing with a crisis in power usage, power cost, or both.
A Gartner group survey found 70 percent of CIOs are concerned about power and cooling. In addition, Gartner expects energy-related IT spending to double in the next five years. That means half your IT hardware investment may be associated with energy, heating, and cooling. If it isn't already, Green IT will be an important topic in 2008.
There's a wealth of information on green or sustainable computing available from many sources. It's difficult to know where to start, but one good approach is to evaluate each option in terms of its immediacy and impact.
Here are four major approaches you can take to 'green' your IT effort. They are not mutually exclusive. You might start with the low-hanging fruit that you can execute on today, and lead up to the more capital-intensive options that deliver longer-term benefits.
Four major approaches to Greener IT
- An easy first step is some basic cooling and system configuration changes. Many organizations are becoming aware that they are running their data centers too cold. Computers can be operated safely 3 to 4 degrees warmer than they are currently in many data centers, and this can be an easy and valuable adjustment to make. Another relatively easy fix is to use the power efficiency features that already come with your systems, either as part of the hardware management or operating system software. Optimizing these settings will reduce energy usage by powering-down and hibernating more often than in the default energy scheme.
- The next most immediate change, and one with a greater payback, is to deploy software solutions enabling server and storage consolidation. This includes software for virtualization, automated storage tiering, data deduplication, clustering, and advanced workload management. This has several benefits. First, it reduces hardware requirements, and the related energy consumption. This in turn reduces the heat in the data center, and the load on your HVAC systems. Finally, you'll realize labor and maintenance cost savings from having less hardware to administer. These projects can be preceded or combined with application rationalization and standardization efforts so that you are maximizing efficiency around fewer, more strategic applications and tool sets.
- The third approach is to upgrade to the latest hardware. This is more costly than the first two options, but can yield even greater results. Newer computers and monitors, especially those with Energy Star ratings, are more energy efficient that older ones. This should also be true for storage. In addition, new classes of storage such as MAID (Massive Array of Idle Disks) are even more energy efficient, and can be used effectively within a data lifecycle management program. Understandably, this approach, while yielding measurable power savings, would need to be implemented within an established technology refresh cycle for most companies.
- Finally you can undertake data center redesign and/or consolidation. Depending on earlier application and server consolidation success, you may be able to reduce the number of data centers you are running. Or you may choose to outsource data center management to a managed services partner who can do it more efficiently on a global scale.
For the data centers you choose to run yourself, it might make sense to hire an engineering firm that can help you improve airflow management, electrical infrastructure and other facility elements. Consultants employ computer fluid dynamics models to optimize chilled air throughput and minimize hot spots. Rack schemes can be changed to accommodate 'hot' and 'cold' aisles, and physical plant changes can be made to address static pressure and condensation. This is typically a multi-year program, relatively expensive, and will most likely incur downtime to the data center operations. Ideally you should combine major data center improvements with physical data center migrations.
All of the approaches discussed here complement and build upon each other. Start down the path to a greener IT effort today, and realize the benefits of power conservation, server and storage efficiency, and sustainable computing.
How software can help you take a lead role within your company to reduce energy requirements
Types of software that reduce energy needs include virtualization, automated storage tiering, data de-duplication, clustering, advanced workload management, and desktop management.
The right software can provide multiple benefits when it comes to reducing energy use as well as costs. First, it can reduce hardware requirements, and the related energy consumption. This in turn reduces the heat in the data center, and the load on your HVAC systems. Finally, you'll realize labor and maintenance cost savings from having less hardware to administer. Let's look at some major categories of software and how they can help you "green" your IT effort.
Classes of Software that deliver Greener IT
- Virtualization: Server Virtualization is a hot topic, thanks to the publicity surrounding one of last year's most successful IPO's, VMware. Virtualization helps you increase server utilization by pooling applications onto fewer physical servers. This allows you to either decommission unused servers, or more typically, defer new server purchases.
- Advanced Clustering: New approaches to system clustering provide high availability with fewer standby servers. In the past, failover protection required a dedicated standby for every operational system, but advanced clustering software now lets you cluster many systems that leverage a single 'roaming spare.' This configuration is known as 'n+1' clustering. Another option is to have multiple systems draw on the resources of multiple backups, even if those systems are already handling some other workload—a configuration called n+m clustering. Either approach allows for server consolidation, increasing overall utilization and reducing hardware and energy needs.
- Dynamic storage tiering: This is an advanced Storage Resource Management capability that lets data move across different tiers of storage in response to data characteristics and environmental criteria. It achieves two important objectives. First, it increases storage utilization by connecting otherwise separate "islands" of storage, improving end-to-end visibility of all storage assets, and identifying unclaimed storage. Second, it shifts data automatically from high-end, operational arrays down to near-store or archival devices. Dynamic storage tiering also allows you to utilize different classes of storage hardware, which yields power savings.
- Data deduplication: Traditional backup methodologies have some companies backing up the same data 20 times or more. Similarly, multiple copies of the same email can end up being stored in many different places, such as personal folders, servers running Microsoft Exchange, file storage, etc. Deduplication software, however, can reduce or eliminate duplicated data along with the need for all the storage hardware to hold it. Using deduplication during backup, along with single instance message storage, can reduce disk space usage and also cut bandwidth needed for backups by 50x–500x.
- Desktop management: This type of software provides better control over desktop power usage than is possible with the power management features typically supplied with operating systems or hardware. Desktop management software simplifies the power-down of idle desktops, and automates power settings. This is especially important across distributed, global enterprises with thousands of desktops and devices.
- Application Performance management: IT organizations often over-provision server and storage hardware to solve performance problems. There's another alternative: by using Application Performance management software, IT administrators can automate the process of identifying and addressing performance bottlenecks. That allows for optimally tuning databases and applications, which can improve performance 30 to 95 percent without adding or upgrading hardware.
Using software is a great approach to greening your IT effort. It can provide fast time-to-value—and a compelling return on investment.
Article: Advanced Cluster Server Solutions in a Nutshell, by Minakshi Sehgal
Article: XML Guide for Dynamic Storage Tiering (DST), by Marianne Lent
Article: Global Data Deduplication, by Peter Elliman and Stefaan Vervaet