The Looming Paradigm Shift In Data Storage

It seems like this would have happened a long time ago. Or, maybe it did, but everybody just accepted the status quo rather than clamoring for a better answer.

The problem? Data storage. The proliferation of the Internet, the computerization of consumer information, and the sheer amount of data available for storage has reached farther than anybody could have ever imagined a decade ago. Take for instance the computer my parents purchased in 1997. The salesmen told them - and was probably being as honest as he could be at the time - their hard drive would be all the storage they'd ever need. The size of the hard drive? Three gigabytes. For comparison, most of today's new computers have hard drives approaching 200 gigabytes worth of storage capacity.

The next ten years could see the same degree of exponential growth in information retention needs, particularly for corporations with massive amounts of data to retain.

The problems with traditional storage methods (the hard drive) has become apparent though - giant hard drives are risky, in addition to being limited in terms of functionality. If a hard drive breaks, the data is just usually lost....something most corporations simply can't afford to risk. Even when the hard drive is working though, most only allow a single point of access, called a node. Therefore, they may be slow or completely pointless when there are dozens or even hundreds of users needing to use the stored information.

The solution to the dilemma appears to be clustered storage...an idea first brought into reality abut six years ago. A clustered file system is a computer file system which simultaneously uses multiple servers. When a database is stored on an unclustered system, the complexity of the underlying storage environment increases as servers are added. After more than a few servers are introduced into the network, the system becomes increasingly complex, and increasingly inaccessible. With a clustered file system, the complexity is avoided, which allows a very large numbers of servers to be clustered together....and added to later. With this approach, every server has simultaneous read and write access to those files. And, multiple users can access the entire system - by scaling the input/output bandwidth. And best of all, all that information isn't at risk if one component fails, even though it operates like a single system.

Too Few Players....For Now

So what public companies have a hand in this budding innovation? Considering the concept is not exactly complicated, there are surprisingly few. The barrier to entry is probably not the hardware, but the fact that software needs to be written to make the hardware work within a clustered data storage architecture. Well, that, and the patent protection many of these early companies already have. Regardless, there are a handful of companies with what we see as substantial promise in this relatively new arena. The problem is, very few are publicly traded....yet.

Perhaps the best known name already with an investable presence is Isilon Technologies (NASDAQ: ISLN), even though the stock has only been trading for a few months. They may swing to a profit this year after several quarters of sales and earnings improvements. Data Domain (NASDAQ: DDUP) is another newcomer, having gone public in June. Though it's a rarity, DDUP shares are actually higher now than they were after the IPO. Brocade Communications (NASDAQ: BRCD) is warming up to the concept as well, and don't forget Mellanox (NASDAQ: MLNX) either.

As for the others, the buzz is there should be several IPOs in the coming months for companies in this niche. A company called Compellent is slated to go public sometime in Q3. Crossroads, Panasas, Continuity Software, Pivot3 and Exanet are just some of the other names that may be experiencing enough demand in the near future to justify going public. Though it's tough to really make any money as an IPO investor - unless you actually get the issue price - this industry's stocks don't seem to be relying on mere hollow hype. So, while they may not be overnight stock sensations, we do think the idea is worth keeping in your back pocket (if only because the majority of the market still doesn't have their finger on this pulse yet).

Grand Finale

It's difficult to say clustered storage is going to be a mega-trend in the same sense John Naisbitt, author of 1982's popular 'Megatrends' book, might use it. We do, however, expect this data storage revolution to be an investment-worthy opportunity. Why? Not because the investment community is saying so, but because the technology folks who make the purchase and upgrade decisions say so.

SearchStorage.com - a site targeting the IT community rather than the investor community - has dozens of articles singing the praises of clustered storage. EWeek.com recently published an article that seemed about as direct to this point as it could possibly be, titled 'Cluster Storage Is Now The Norm'. Straight from the article....."Six years ago, as cluster storage was just beginning to ramp up in production use, conventional, non-linked storage systems constituted about 80 percent of the HPC market. That has all changed dramatically. Cluster systems took a majority (about 52 percent) of the market in the first quarter of 2006 and haven't looked back, according to research company IDC. Storage clusters are now comfortably past the 60 percent mark in market share, researchers reported."

Don't confuse their comments with total utilization, as it might seem as if the total market is already saturated with clustered storage systems. They're referring to the 'new' storage market. There's a plethora of storage upgrades on the horizon - enough to fuel growth for many of the players in the industry.