FORGET ABOUT automated information lifecycle management (ILM). Before one Fortune 500 biosciences firm ventures into ILM or tiered storage, end users will first have to classify all of the files they've generated over the years, says Michael Masterson, information systems architect at the firm. Without a proper data classification, he says, "ILM is putting the cart before the horse.

"IT can't classify those files," adds Masterson. "They're not the information owners." To that end, the firm is training users on Abrevity FileData Manager, which supplies them with an interface for tagging files similar to the way in which Apple's iTunes lets you create playlists. Once data is tagged appropriately, "ILM is downhill," he notes, and just a simple question of data movement.

The Abrevity tool is the first one Masterson has seen that can capture a file's context. "It's unique in its ability to capture [file] attributes in a loosely structured way," he says, including the information you can gather about a file based on which folder it's in. "Somebody has already automatically classified that file by putting it in a certain folder." Other data classification tools Masterson considered look at coarse file meta data or rely on extensive ke...

Abrevity's ability to contextualize files is related to its unique SliceBase data model and not a generic SQL database. "With SQL," explains Masterson, "a designer has to come in and predefine the columns. It doesn't have the flexibility to just go out there and find whatever is there."

Masterson's company has good reason to embark on a data classification effort. As a public company developing medical equipment, the firm is regulated by both the SEC and FDA. "That's before we get to the classic ILM cost-of-infrastructure concerns," he says.

Luckily, the data volumes Masterson's division needs to classify are relatively small, so they don't need to implement a nearline tier of storage right away. "We have under 10TB of free files, so we have the luxury of perhaps using more disk than we need to," he notes.

Data classification has become a hot topic as of late, and a number of startups are jockeying for attention. Other vendors to offer some form of data classification include Index Engines, Holmdel, NJ; Kazeon, Mountain View, CA; London-based Njini; Scentric, Alpharetta, GA; and StoredIQ, Austin, TX. Long term, analysts expect server and storage vendors to license data classification software directly.

--Alex Barrett