Today, many customers strive to achieve environmental leadership while regulating costs across their business activities. Reducing storage hardware costs, cutting energy consumption and meeting IT demands is a battle that is raging in almost every data center in the world. The EPA recently found that U.S, commercial electrical costs increased by 10 percent from 2005 to 2006. In order to cut storage costs, maximize energy efficiency and create a greener data center environment, customers need to employ a green storage strategy. They need to invest in initiatives that diagnose their storage infrastructure, determine the requirements for data usage, establish plans to use storage efficiently, monitor their actual storage operation, and include storage in their plans for efficient data center power and cooling:
Step #1: Diagnose your current storage operation
The good news for data center operators is that there is already considerable help available in technology and best practices for storage use. With the help of software tools, the storage can be used very effectively. Many users will find that they can lower both the operational and capital expense by improved management, which starts with an understanding of requirements for stored data, and a comparison to current actual use.
The first step is to understand the requirements that enterprise data has for access, performance, and protection through local and remote copies. For most data, these requirements will change as the data ages, or moves through its lifecycle. There is considerable variability in the power consumption of different storage types, with more power required for better performance or higher levels of protection. Understanding the actual needs of data allows the correct protection for valuable data, without overspending on data that could be stored on cheaper media.
Data classification includes understanding what data should be discarded, or relegated to a low power shelf media like tape. It can also help users understand that they may be storing data that they shouldn’t, like a personal MP3 collection on a high performance SAN.
Finally, there is a need to understand how storage capacity is allocated and used. Historically, mainframe users have been able to achieve capacity utilization rates of 70 percent or better. Through a combination of inadequate tools and inattention, users have often fallen far short of that mark in Unix, Windows, and Linux systems. By improving capacity utilization, often possible by a factor of 2, users can lower their current power bills and defer capacity upgrades.
Step #2: Build the right storage infrastructure
With a clear understanding of storage requirements, an efficient infrastructure can be built. The deployment plan should include a mix of high performance disk arrays with FC or SAS drives, and tier 2 arrays with SATA drives, as there is a considerable difference between the power requirements for these types. Arrays with space efficient flash copies may be preferred, as these can provide copies for backup processing with minimal use of physical capacity. Tape as a high capacity and low power shelf media will be a key component of most enterprise configurations.
With different types of storage playing roles in the infrastructure, placement of data on the correct storage type is critical. Information Lifecycle Management (ILM) software is used to move data within a storage hierarchy depending on its usage. IBM has found that using a storage virtualization layer under ILM allows the creation of storage pools with different performance and power characteristics, and the easy movement of data between pools.
Technology continues to improve storage efficiency. Offline (archive) facilities are now being built with a data de-duplication function that eliminates redundant data and decreases the need for physical capacity.
Step #3: Monitor operations
Even with the right diagnosis, hardware and ILM solutions in place, a data center can only be as efficient at its management, and a well managed storage infrastructure will maintain energy efficiency. Data center resources are typically assigned to manage peak workload conditions, but most data centers run well below peak capacity the majority of the time, with hardware consuming power even when not being utilized. By implementing automated storage management software organizations realize dramatic cost savings. Additionally, utilizing real-time software tools to monitor actual power draw can help modify power consumption. Capacity planning and implementing storage management software to move infrequently used data from high performance disk to low performance tape storage should also be taken into account.
Step #4: Cool things down
With the amount of hardware residing in a typical data center, cooling becomes one of the most critical systems necessary to maintain a data center, and it has been estimated that each year organizations pay a significant part of the purchase price of a data center on powering, cooling, and housing the systems. Without proper cooling, a data center running on backup generators during a power outage will shut down in minutes. Conducting regular thermal assessments and exploiting liquid cooling solutions when feasible--inside and out of the data center, will dramatically reduce and maintain data center costs.
Diagnosing, building managed infrastructure, monitoring, and cooling are integral to the business goal of maintaining storage costs while simultaneously reducing energy consumption in the data center.
By following these guidelines, IT managers can dramatically reduce data center consumption and begin creating a greener storage environment for their organization.
Clod Barrera is the Chief Technology Officer and Distinguished Engineer for the IBM System Storage business. Clod is also the co-chair of the Storage Networking Industry Association’s recently launched Green Storage Task Force.