Home CTR Exclusives Getting Ahead of the Data Storage Energy Crisis: The Case for MAID

Getting Ahead of the Data Storage Energy Crisis: The Case for MAID

There is a tremendous amount of focus on energy efficiency these days. Some think it’s a nice “green” thing. Others understand the significant financial benefits that can be gained. But for those of us in the storage and networking industry, it’s a lot more urgent than that. We are on the brink of an energy crisis that fundamentally impacts all of our professional lives.

Consider this:

  • Data centers are now responsible for more than 1.5 percent  of energy consumption each year in the entire U.S., according to a recent EPA report.
  • IDC reports that annual data center electricity costs exceed $3.3 billion.
  • The cost to cool a data center now exceeds the cost to lease the space for it. For example, a 100,000 square foot data center can cost $6 million per year in utilities alone by some estimates.
  • Gartner predicts that in five years, energy costs will consume up to 40 percent  of IT budgets.
  • By 2011, 97 percent  of data centers expect to be out of power (just two years away!), according to Liebert, a power supply company.
  • Many government data centers are already out of power and cooling resources

The situation is so serious that the government is starting to mandate energy efficiency. Back in 2006, Congress required federal government data centers to install energy-efficient servers. President Bush recently signed an executive order requiring federal organizations to find ways to save energy. Overall, the Fed has set a goal of 30 percent reduction in energy consumption by the year 2015. The current economic crisis and changing political environment could certainly accelerate efforts like these and expand them to other industries as well.

Data Storage – Guilty!

As we all know, data storage is one of the main culprits. Data is mounting quickly. According to ESG, a 50 percent plus annual growth rate is commonplace -- and many organizations report rates well in excess of that, especially in government agencies. The explosion in data storage needs has fueled dramatic increases in our storage hardware footprints and has more than offset the energy efficiency benefits of server consolidation realized from virtualization technologies.

Today, 40 percent  of power in the data center is now for data storage, according to the InfoPro Wave 9 study. Why? We simply don’t delete data anymore. By many estimates, 75 percent  or more of the data in a data center is retained for longer periods of time. This is being driven by corporate and legal compliance requirements and the need for data accessibility for future re-use. These requirements affect nearly every industry and business segment including manufacturing, retail and service, financial, healthcare and government. As a result, most of the usable data remains -- and is expected to remain -- on disk. And that means storage-related power consumption will continue to increase over time.

In the short-term, some storage vendors have decided to replace existing lower capacity Fibre Channel drives with higher capacity Fibre Channel drives. While this may equate to incremental power and cooling savings, it does not solve the inherent issue. These disks, although higher capacity, are still ‘powered on’ 100 percent of the time, which does very little to stem the tide of explosive persistent data growth, and the increasing energy needs in the face of finite data center power and cooling capacity. In addition, replacing lower capacity drives with higher capacity drives introduces performance degradation.

Using higher capacity drives to reduce energy efficiency is analogous to replacing 60-watt incandescent bulbs with 25-watt fluorescent bulbs in a building. However, if the building is empty, why keep the lights on in the first place?

Persistent Data vs. Transactional Data

Data center managers are starting to recognize that 70 percent to 90 percent of their data is not transactional in nature, but rather, persistent data that is rarely changed, and infrequently accessed. The quality of service (QoS) requirements for persistent data are less demanding in terms of system performance (I/Os per second), access time latency, and levels of concurrent data access. Combine the lower QoS levels required for persistent data with the other storage system characteristics, which are optimal for persistent data, including dependable long-term data retention, higher reliability, higher data integrity, low power and cooling, and small footprint, and you have something that can start to address the data storage energy crisis. 

The key to managing persistent data is to deploy storage systems that are tuned specifically to its needs. One technology that is ideally suited for persistent data is MAID (Massive Array of Idle Disks) storage. Smaller footprint MAID-based systems reduce disk power usage by powering up disks only as needed, and limiting the maximum number of disks that are in use at any one time. Because unneeded disks are idle or powered down, MAID technology consumes much less energy -- not only to run disks, but also to cool them. Limiting power consumption reduces data center energy costs and increases the effective life of the disks themselves. This increases overall disk reliability and enhances data protection. All of these attributes are a perfect match with persistent data accessibility requirements.

Some MAID technologies also have the ability to control where data is written and how it is accessed. This added intelligence enables users to power drives off completely (instead of just spinning them down). That allows for denser packaging of drives and more power-efficient array shelves -- saving additional power as well as data center floor space.  

Energy and $$$ Savings 

The latest MAID technologies introduce a dramatically new level of storage system energy efficiency, and coupled with the industry’s highest single footprint capacity, provide cost savings ranging from 75 percent  to 90 percent of energy per unit of storage, and a 5x reduction in footprint. The resulting economics are extremely compelling with an estimated four-year power and cooling savings between $9.5 million and $13 million, as compared to data stored on traditional disk array solutions. However, perhaps even more compelling than the economics, is that data centers may not even have the power available to store and manage their data “as usual.”

Where power and cooling considerations were once restricted to sizing a UPS, they are now fundamental limitations in how a data center can grow to meet future business demands. When planning for storage growth, data center managers must now consider the environmental impact of any equipment to be installed. Traditional storage technologies do not address this problem. To avert an energy crisis in your data center: identify your growing persistent data and move it to smarter, more energy-efficient storage systems.

Will Layton is co-founder of storage solutions provider COPAN Systems.

 

 

Information Technology Jobs
Keywords:
Location:
Job category: