Shrinking for growth: a unique storage solution that reduces energy costs

On the energy front, California has clearly had it rough. Deregulation hit the state hard, leading first to power shortages and rolling blackouts that dominated the national news, and then to a lot of finger pointing and blame.

Part of the problem, of course, was the complicated set of events that the 1996 deregulation bill set into play; but a sizeable portion was also due to continually rising energy demands. What was good news in some sectors – strong economic recovery meant tech companies were thriving and growing – led to tough news in others, as the grid struggled to keep up.

But as a recent CNN article pointed out, the Golden State wasn’t alone when it came to ever-increasing demand for energy, and for years, tech growth has threatened to outstrip energy supplies throughout the United States. “Increased use of computers and other appliances and the lack of adequate new power plants are the norm across the country,” the article’s authors warned. In other words, California’s woes were just the beginning.

Now California is leading the way again – but this time on a more positive note: One of the state’s largest utilities, Pacific Gas and Electric (PG&E), has begun offering incentives to companies that deploy energy-efficient storage systems based on a technology called Enterprise MAID (Massive Array of Idle Disks), developed by Colorado-based COPAN Systems. Having recognized that data centers are among the most power-demanding elements in many of today’s organizations, PG&E has stepped up to help them solve the long-term problem of runaway energy needs.

The Hungry Data Center
The need for more efficiency in the data center has become an issue in many companies as data processing and storage volumes have increased – sometimes sharply – over time. Energy prices have brought the dilemma into sharp relief, but a number of factors have been pressing the problem for years.

Certainly data processing and especially persistent data (write once read infrequently) is on the rise. Companies that once had a few internal systems to worry about now have dozens of applications constantly running, and networks host traffic all day long, carrying everything from memos and spreadsheets to inventory and order processing.

Data stores have also grown enormously. Not only does organic business growth mean there are constantly more records to be kept, but new regulations call for longer-term retention for many of them. Government-enforced policies and companies’ own rules affect nearly every industry and business segment, and some – such as healthcare and finance – are being hit particularly hard. In addition, the persistent data that’s stored often needs to be kept on a highly accessible system so that legal requests and other demands can be met without delay. Altogether, that means more information is being kept on disk – with its inherent power and cooling needs – and less is being off-loaded to static systems, such as tape.

Disaster recovery has compounded data center needs too, as highly publicized tape transportation losses and other catastrophes have spurred more organizations to replicate their data offsite. And even the increasing demands of global competition call for high-speed, constant access, which often means more power-hungry systems are required.

Of course, all these increasing demands are occurring in an environment of flat or declining budgets, and many companies have added energy conservation goals to the CTO’s plate, on top of that do-more-with-less scenario. All in all, calling the current atmosphere in the average data center a crisis – or at the very least, an urgent situation – is not an exaggeration… it’s a reality!

With the increase in persistent data storage alone, several companies have projected that they’ll outgrow their current data centers in a matter of years, and some say they’ve considered building additional facilities to house the overflow. The cost to those businesses would be substantial, of course – but even more worrisome could be the effect that thousands of runaway data centers would have on overall power consumption.

Storage as a Solution
Of all the factors affecting data center growth, the primary culprit can be traced to one simple fact: Most of the data that organizations generate – from important patient records to long-forgotten memos – is never deleted from the system i.e. it’s persistent. By most estimates, 75% of a company’s data is kept perpetually, which means the growth trajectory is steep. According to a 2006 study by The InfoPro, disk-based storage-area network capacity (most of which is persistent) in the average Fortune 1000 company increased from 138 terabytes in early 2004 to more than 680 TB in October of 2006, and it has doubled every 10 months since 2005. At that rate, it’s expected to exceed 8 petabytes in only four years.

A factor that exacerbates the challenge is that an increasing percentage of that persistent data, for either regulatory or competitive reasons, needs to be kept on-line i.e. on a disk based platform.

Keeping data accessible means maintaining it on disk-based storage, and traditionally, disk storage has been expensive to operate. Spinning spindles draw power and generate heat, which means they take energy to run and more energy to cool down. The higher the system’s density, the more disks that are packed into a small space, and the more cooling that’s generally required. Of course, that expense has to be weighed against the organization’s need for fast access to the information, and for transactional data – the information that’s being created, accessed and changed day-to-day – spinning disks are still the best solution.

But data center managers are realizing that the vast majority of the data they’re now required to keep – 70% to 90% of it, in fact – isn’t transactional in nature. Instead, it’s persistent data that’s rarely changed and infrequently accessed. When persistent data is needed, its quality-of-service (QoS) requirements are less demanding than those of transactional data, which means I/Os per second, access time latency, and levels of concurrent access all come with less stringent requirements.

That’s good news from an energy and power point of view, because it means most of an organization’s data can be stored on a system with lower energy demands. And that’s where COPAN Systems solutions comes in.

Idling the Disks
With COPAN Systems’ Enterprise MAID technology, a “massive array of idle disks” is specifically configured to hold an organization’s large stores of persistent data. Enterprise MAID operates on the principle that, for long-term storage, a system’s disks don’t need to be kept constantly spinning. Instead, a maximum of 25% of the drives are powered on at any one time.

The energy savings are obvious, from both a drive-powering and a system-cooling point of view. In fact, not only does an Enterprise MAID system generate much less heat than a traditionally, always-spinning array, it also takes up much less space in the data center, which compounds the resource advantages. And because the disks spend less time “working,” their drive life increases and they experience less downtime. That increased reliability, of course, means better data protection.

The primary advantage of Enterprise MAID, though, is its extremely high capacity per unit of power. COPAN’s systems have been shown to have a power density of more than 90 terabytes per kilowatt hour, which compares favorably to the average Fibre Channel disk product (4 TB/KW) as well as the average SATA disk product (17 TB/KW). These efficiencies translate into enormous power and cost savings, and of course the larger the data stores, the greater the efficiency.

Reducing Waste Even More
When PG&E decided to offer incentives to companies installing COPAN Systems’ Enterprise MAID platform, they based their assessment on the power-down factor alone, recognizing that an array that’s spinning only 25% of the disks will obviously result in enormous energy savings.

But the economics of data storage are more complicated than that, and any organization looking at long-term storage today needs to consider a number of factors: not only accessibility, footprint density, and power and cooling costs, but also data growth, WAN-usage levels, business continuity requirements and more. While the hardware innovation of Enterprise MAID storage has made it economically viable to store persistent data on disk instead of tape, additional technologies have made it possible to increase efficiency even more. Specifically, de-duplication is an important software-based advance that lets organizations trim waste out of their storage systems by drastically reducing multiple copies of the exact same data.

Because a company’s backups accumulate a lot of redundant data over time – some of it duplicated 10 to 30 times – creating the most efficient storage system possible means cutting that waste, as well – and well-planned de-duplication efforts can reduce a bloated data store by a factor of 10:1, 20:1, or even more.

The most obvious effect of decreasing the amount of data stored is the reduction in the amount of hardware required; less data calls for fewer disks. Of course, the cost savings of smaller systems are obvious at purchase time, additionally less hardware requires less power, as well, so the economies compound over time. In addition, smaller data sets are faster to transmit, leading to a positive impact on both replication and disaster recovery. For many organizations, that means data that was once backed up on physical media and trucked offsite can now be sent over a broadband network.

Helping Business Shrink Energy Consumption
In these energy-conscious times, many organizations have set aggressive goals to reduce waste, trim costs, and consider the most efficient use of every resource they have. And with the current worldwide focus on the environment, many have developed energy-conservation objectives, as well. But even in the face of rising costs, uncertain supply, and the need to fulfill ecological goals, it’s not always easy for businesses to make the leap to systems with a long-term payoff.

Like many utility companies, PG&E has seen the synergy and the benefits of offering incentives to businesses that change their operations to minimize waste and reduce consumption. For organizations that break out of the traditional storage box and adopt COPAN Systems’ Enterprise MAID technology, they see short-term savings as well as long-term payoff.

PG&E says it will continue to expand its program to encourage conservation and help organizations make positive environmental and business decisions – and of course, many forward-thinking utilities are expected to follow.

Staving off another energy crunch is a goal that takes plenty of cooperation – between businesses, utilities, equipment manufacturers, and more. With a long-term focus on constant improvement and renewed interest in environmental concerns, many technology and other storage-intensive companies are prime candidates to make changes. And utilities are prime candidates to help.