Tiers Without Tears

Tiered storage, the art of assigning data to progressively cheaper devices based on its age or relevance to an organization, is getting the hype treatment deluxe these days. But the actualities of tiered storage are proving to be a challenge for even the most seasoned IT pros. That said, those who can successfully put tiered plans in place are apt to ensure their employment longevity.

There's clearly a demand for tiered storage architectures. Companies are inundated with unprecedented levels of data, much of which requires special handling. And while the cost of disk is dropping, the hardware, software, and support surrounding disk solutions are not. Hence, it only makes sense to start ensuring that data isn't just organized, but that it's organized in ways that make the most efficient use of resources.

"All of us are ... looking at how we retain and categorize documents and decide how long they live. Everybody has to look as some tiered storage," says Bob Mason, CTO of Spanish-language newspaper firm ImpreMedia and former director of publishing systems at The Dallas Morning News. (See Dallas Morning News.)

The problems come with implementation. Storage pros need to answer lots of questions up front: Just how will data be organized? Is it sufficient to use metadata parameters, like the date of last access, to set up an automated policy? Can an automated policy be used at all? How many tiers should be set up? What kind of storage will be used for each tier?

The key element in getting started is knowing what data you have and what you want to do with it. Potential court documents, for instance, must be stored with clear timestamps and "chain of custody" records. (See Storage Goes to Law School and Lawyers Urge Doc Management.) For compliance, the information the government wants must be relegated to storage that's easily accessible. (See Top Tips for Compliance .)The pundits recommend keeping it simple, but that's easier said than done. It may be helpful to classify data according to military-style grades of accessibility, for instance, but what about email, which wasn't necessarily a factor in past data classification schemes and may not even be stored in the same geographical location as other "top secret" documents?

Then there's the issue of executive buy-in. Without the cooperation of key managers, any kind of tiered storage is likely to topple before it gets built. What self-respecting manager would agree to have documents put on cheap tape when other departments are enjoying disk? How can IT adequately represent the need for prioritizing and tiering without getting into political hassles?

Products can help, to some extent, by automatically setting up classifications that can be tweaked and modified. But there is no magic bullet. As analyst Dan Tanner of the ProgresSmart consultancy told us last week, "There are lots of tiered storage products, but you must do the data classification yourself or use a separate package for that. Or, you can use data classification products and pick your own tiered storage." If you're an SMB, there are a couple of products, including the new multivendor ATS offering, that combine several elements. (See Breece Hill & Pals Unveil App.)

Despite the difficulties, it seems wise to at least get started, since the challenges of tiered storage will only strengthen as time passes. Even a choice to relegate "old" data to VTL is a step in the right direction.

Can tiered storage be implemented without tears? That's unlikely, at least among businesses that run on the capitalistic model, which emphasize internal as well as external competition.Still, it is likely that in the years to come, the job of storage manager will consist largely of designing and assigning specific living arrangements for the various data stores inside a particular organization. If a manager can do that successfully, his or her secondary function will be to bring in the best technology to support the classification.

Mary Jander, Site Editor, Byte and Switch