Green storage technology reduces data center power consumption

By Suresh Nair, Managing Director, NetApp ASEAN | Feb 20, 2009

In the current economic climate like it or not - the march toward green technology will keep going and so will the mantra of doing more with less. For some organizations, moving to energy-efficient storage isn't just a nice, green idea but an essential part of using less power, which in some circumstances has become an extremely limited resource.

The wide array of technologies that fall under the green storage heading can all be used to rein in the data center's fastest-growing overheads, and can help postpone the expansion or relocation of data center premises. In fact, it may even be the only way for companies to manage the continual growth of data, and thereby stay in business especially during these times of recession.

The data center - with its escalating demand for power, cooling and space - is increasingly seen as one of the major culprits generating an estimated 170 tonnes of carbon dioxide emissions, or 0.3% of the global total.

As corporations continue to grapple with escalating storage requirements, disaster recovery protocols and globalization, this will only get worse. A recent McKinsey report predicted that data center emissions will overtake those of the world’s airlines by 2020. Not to mention that data centers now account for 25% of the IT budget, with energy bills growing at 16% a year.

Exponential data growth is an undeniable reality for most data centers. IDC Worldwide Disk Storage Systems Forecast 2006–2010 predicts that worldwide data will grow at a compound annual growth rate of 50.6% through the decade. This growth is a huge concern for IT managers.

Until recently, continuous improvements in price, performance and $/GB have made it both easy and affordable to solve storage concerns simply by adding more disks to existing storage systems.

However, IT executives are discovering that there are limits to that easy growth: floor space, weight loads, rack space, network drops, power connections, cooling infrastructure, and even power itself are finite resources. Hitting any one of these limits significantly jeopardizes the ability of IT to meet the demands of business.

Hitting the fan
Cooling is inextricably linked to power consumption. Every watt of power that enters the data center generates heat that must be removed from the environment - and to do so takes more power.

As the heat increases, systems become more unstable and component failure rates rise. The cost of power to cool a system often is as much as powering the system itself. But cost is not the only issue. In some cases, rack densities generate more heat than existing cooling infrastructures can handle. Where there is no room for additional cooling infrastructure, there is no room for growth.

Power hungry
Companies have been steadily increasing storage density, inadvertently driving up data center power demands and cooling concerns.

In the next 18 months, increases in average storage rack density are expected to drive average power consumption from 2kW per rack to 30kW per rack - and that’s only half the story. For each watt used by the server or storage, A/C, power supplies, and other related equipment together require nearly 1.5 times that amount.

Putting business at risk
The combined effect of recent increases in the price of energy and the adoption of denser computing and storage architectures has driven energy costs for some data centers to 30% of their total operating budgets. If left unchecked, the cost to power IT equipment could exceed its acquisition cost in a matter of years.

Without deliberate action, high energy costs will cripple an IT department’s ability to grow and change in support of the demands of the business.

 

The greening of storage
Environmental issues are gaining serious commercial momentum and, fueled by the growing number of local and global green initiatives, they are rising ever more insistently up the corporate agenda.

More power-efficient storage solutions provide for business growth while saving power. Every watt of energy saved in the data center is a watt that is removed from an organization’s carbon footprint and the global warming equation.

So how can users combat these power and cooling issues in the face of growing storage demands? Many energy-saving ideas have been professed by academics and researchers, from throttling workloads, liquid cooling, to mechanisms for powering down components of a disk drive.

In particular, a number of vendors have been promoting the idea of turning off disk drives during idle periods.

Although promising in concept, this technology is highly complex and may require delicate tuning to achieve true energy savings. For example, difficulty in accurately predicting idle periods can result in disks spinning up soon after they were spun down, resulting in less energy conservation than anticipated.

Furthermore, there may be risks that drive mechanics will become less reliable with repeated spinning up and spinning down. In the absence of conclusive evidence that proves the long-term feasibility of turning off disk drives, customers are hesitant to adopt such technologies in haste.

A way out
The approach to fighting rapidly growing power consumption is simple: subtract machines and disks from the power equation by using storage more efficiently.

This strategy has many corollary benefits - it lowers complexity, lowers people costs, lowers support and service costs, and improves network efficiency and performance through the following 7 steps:

1. Consolidation of servers and storage
Servers alone can consume 50% of the power coming into the data center. The first step in reducing power consumption is to attack the power problem where you can reap the most gains - consolidating and virtualizing application servers.

Storage is the next largest consumer of energy after servers and cooling systems. In environments with lots of direct-attached storage, as much as 27% of the power going into the data center is being consumed by storage.

The dramatic growth of Windows data has led to a proliferation of Windows file servers and direct-attached storage units. Not only are these server and storage assets poorly utilized, they are also difficult to manage, resulting in extremely high overhead costs.

Data centers that use lots of direct-attached storage can see significant power savings by implementing a storage network. Removing file servers immediately increases the total watts available in the data center.

File server consolidation also increases your ability to scale in a power-efficient manner; instead of adding servers, you can scale by adding disks. Adding disks to an existing system has the additional advantage of
allowing you to amortize operational costs over more drives.

2. Usage of higher-capacity drives
Using higher-capacity drives can significantly alter the storage power equation. Typical SATA disk drives consume about 50% less power per TB than equivalent-capacity Fibre Channel drives. They also offer the highest available storage density per drive, further helping minimize power consumption.

NetApp disk resiliency and data protection technologies have made SATA drives an increasingly viable alternative for many enterprise applications. Many companies are finding that SATA performance is good enough to support many more applications than they originally thought possible.

3. Protection against disk failures with fewer drives
Disk drive arrays may provide several advantages over the single drive: higher reliability and higher data transfer rate.

By replacing one drive with a group of drives will not increase reliability since the life of the entire system will depend on any of this drives. In fact, reliability - mean time before failure (MTBF) - could even decrease with increasing number of drives since the likelihood for one of them to fail will grow. Thus, a certain level of redundancy is necessary in the design of drive array to increase the reliability of the entire storage system.

RAID is an assembly of disk drives known as disk array which operates as one storage unit. In general, the drives could be any storage system with random data access, such as magnetic hard drives, optical storage, magnetic tapes, etc. When the speed (data transfer rate) is an issue, the fastest SCSI hard drives are typically used.

RAID Functions Redundant Array of Independent Disks (RAID) technology serves the following functions: Immediate availability of data and, depending on the RAID level, recovery of lost data. Redundancy of data at a chosen level. Depending on the level of RAID that is being used, this method of storage provides the redundancy needed for a highly secure system, with additional benefit of faster retrieval of data through multiple channel access. If one these drives fail, they can be normally exchanged without interruption of normal system operation and ensure that no data is lost if one disk drive in the array fails.

 

4. Migration of data to more efficient storage
To ensure the most efficient use of your storage resources, minimize the use of primary storage by migrating data to more efficient secondary storage where appropriate.

5. Increasing storage utilization
According to industry estimates, storage utilization rates at an average of 25-40%. That means 60-75% of all storage capacity that is being powered goes unused. Not only is such a low utilization rate a waste of storage, it is a waste of power.

In most systems, storage administrators allocate and dedicate storage space to a particular volume or LUN at the time of its creation. This creates two significant administrative challenges: (1) once a volume is created on physical storage, its size is extremely difficult to change, and (2) once storage is allocated to a particular application, it is not available for another use.

Since it is difficult to predict actual storage requirements, application administrators typically request much more space than they think they will need to protect themselves should they need more storage down the line. This common practice guarantees over-allocation.

Newer developments in storage technology enables thin provisioning, a technique that lets storage administrators quickly and dynamically resize flexible volumes, eliminating the need for over-allocation. With NetApp Flex Vol thin provisioning technology, utilization rates of 60% are the average.

By making all disks available to all data-sets through a common pool of storage, both performance and capacity utilization are maximized. When disk space is no longer needed by a particular application, it can be returned to the free pool and made available to other applications as their storage needs grow. Increasing utilization 50% results in a corresponding reduction in the number of disks that are required.

6. Backup: doing more with less
Countless copies of data files consume vast amounts of storage. Deduplication technology helps to dramatically reduce the amount of disk storage needed to retain and protect enterprise data.

By identifying redundant files and data as they are being stored, backing up with dedupe allows for a storage footprint that is 50% smaller, on average, than the original data set. Backup data can then be efficiently replicated and retrieved over existing networks for streamlined disaster recovery and consolidated tape operations.

7. Measuring your power efficiency
Our last step is something that we encourage you to do on an ongoing basis to get a handle on your storage power consumption: measure the power efficiency of your storage systems. One common measure of power consumption is watts per terabyte (W/TB).

This measure can be misleading, however, when comparing machines that operate at different efficiency levels. A better way of evaluating storage power consumption is to measure watts per usable terabyte. This can be expressed with the following formula:

          Watts to power system              
-------------------------------------------------------------   =  Watts per Usable TB
Total system TB x % System Utilization

To calculate the power efficiency of a particular storage system, divide the total watts per system by the total number of TBs in that system times the system utilization. System utilization is equal to the percent of your disks that are actually available for use. The figure below shows how this calculation reveals important differences between seemingly similar systems.

Looking ahead
Recession or no recession, the most economically bulletproof sector of the IT business is data storage and its immediately peripheral subsectors, including disaster recovery, e-discovery and deduplication software.

We believe that following the steps that we’ve mentioned above will return enough power and headroom for growth to your data center and protect the ability of your storage infrastructure to support the demands of your business for many years to come.

Will this continue in 2009? Industry experts are near-unanimous in their assessment: Yes, the flow of data will continue to increase, despite a general slowdown in business, and yes, storage and storage-related IT products and services will continue to remain in demand for the near and long term.

Trends in the data storage business in 2009 are pretty well defined. Here are the key ones:

• Virtualized storage and any other IT product that reduces costs will be in demand.
• Automation of virtualized resources will become more strategic to the success of enterprises. "Efficiency" is the operative term here.
• Green IT data center strategies will continue to be deployed -- spurred first by cost savings, and secondly by environmental purpose.

Here are also some of the power-saving technologies that are being investigated by our industry: larger SATA drives, in-line hardware data compression, file deduplication, flash memory, improved power supply efficiency, energy-efficient CPUs, DC power, and intelligent control of the speed of individual drives in response to demand.

Just as today there is no single solution for reducing storage power consumption, future reductions will be attained through a combination of efforts and by attacking the issue on all fronts.