In-Depth

Storage Pain: The Consumer View (Part 2 of 2)

The mints get passed as the conversation turns to information lifecycle management.

• By Jon William Toigo
• 08/30/2005

As you'll recall from last week, I was recounting what I had learned from a “moveable feast” that Computer Associates is currently hosting in cities throughout the U.S. and Canada. I have been participating in this series of free dinner meetings that are providing (in addition to great food) a unique opportunity for IT and business decision-makers and practitioners to compare notes about their storage pains without being hammered by vendor sales or marketing pitches. To their credit, the CA representatives in attendance are doing a good job of remaining silent and letting consumers voice their problems and concerns about data storage. It’s a win-win for consumers and the vendor, where sales personnel are receiving a priceless education in the real-world issues they try to address with their wares.

So successful is the program, which is seeing increasingly greater attendance at each venue (more than 50 attended events in Miami and San Juan respectively), that the Islandia, NY-based company is extending the program to more cities than originally planned. Go to http://www3.ca.com/events/eventdetail.aspx?CevID=72282 if you want to learn more and to register for a event near you.

One of the major topics addressed has been backup pain, which I wrote about in last week’s column. This time, I'll focus on data management and information lifecycle management (ILM).

ILM has not come up as a topic by itself, but as a sidebar—mainly in connection with discussions about data protection, retention, and regulatory compliance. The segue is easy enough to follow: maybe tape backup strategies are being stretched to the breaking point because we are trying to back up everything that anyone has ever stored to disk. Maybe attacking this problem effectively requires that we deal with the real root of the issue: unmanaged, unsorted, and unclassified data.

When ILM comes up in dinner-table conversations, a lot of IT people roll their eyes. One or two state that they see adding more storage as the only effective solution to the problem. Says one fellow, "You just add a tier of SATA disk, with its high capacity with low acquisition cost, and position it as a tape surrogate. That way, backups to tape can be handled outside normal production schedules." Another posits the option of “infinite disk”—just add so much disk that you can achieve redundancy without turning to tape at all.

Mostly, such ideas are shouted down quickly. I ask whether just adding disk only contributes to complexity and labor cost. Others agree: increased disk capacity doesn’t help control data growth. “It lets us get sloppier,” suggests one wise fellow, nodding.

“Not if you have ILM,” offers another, who recounts that EMC was just in his office and a big purchase of the vendor's wares is planned that promises to deliver “ILM in a can.” When I ask him what he means by ILM, he says Documentum, EMC’s recently-acquired document management software product. His company plans to replace a sizeable portion of its Microsoft productivity software on PCs and laptops with workflow templates accessed through a browser-based interface. That way, his end-user files will become neatly organized and manageable.

I note that many companies have opportunities to impose workflow-based management methods like document- or content-management systems across work documents created as a matter of procedural routine. So there are opportunities to rationalize and manage some of the file growth out there with such wares, to be sure.

But, I wonder aloud, does anyone really believe that sales and marketing people are going to go along with such a strategy? Imagine a road warrior working in a hotel room with a lousy Internet hook-up. He needs to prepare a proposal or respond to an RFP. Microsoft Office is at his fingertips, while access to a templated screen of some workflow or document management system hosted at HQ is not. Will he, or the company, be well served by such a strategy?

The files generated by the creative types of the company tend to be the ones that are very difficult to manage, mainly because they are distributed and in the control of the creative workers themselves—each of whom has his or her own data-naming scheme. It is extraordinarily difficult to take away the freedom that has become the mainstay of this “data democracy” and to force the creative worker to march in lockstep with EMC’s (or anyone else’s) notion of document-management nirvana.

Backpedaling, the gentleman offers that maybe certain employees or departments will be exempted from the document management system. Perhaps, their files can be placed into their own file folders or document containers.

I offer that, rather than sorting through and eliminating the junk drawer that is most file repositories today, the proposed strategy sounds a lot like we are using directory systems to build more junk drawers—one for each creative person in the company. Such folders are just as likely to become filled with MP3 files and partially-downloaded bit torrents of the latest film as with honest-to-goodness work-related files.

A central problem with just about any file management approach is its inherent labor cost. Who will examine the contents of the folders, or of the document management system for that matter, to ensure that only valid files are being captured? Without a cadre of overseers, how will you know whether files are being correctly categorized and filed?

I worry that a lot of the ILM/electronic content management vendors are proposing strategies that simply rearrange deck chairs on the Titanic. Like adding more disk capacity, adding lots of file folders gives us the opportunity to grow even more sloppy about data management.

How, then, should we approach data management? Does CA have the solution, someone asks? A lot of suspicious eyes turn to me, awaiting my answer and prepared to pounce if I sound like a paid shill for the host of the event.

My response is quick—and maybe not exactly what CA was hoping for. I say that I was originally drawn to CA because they, alone among the big names in storage, did not claim to have an ILM solution. At one point, late last year, they dabbled with the idea, seeking to capitalize (like everyone else in the industry) on the millions of dollars that EMC was spending to fuel the ILM hype machine. Then (and I don’t pretend to know why) they suddenly removed such rhetoric from their marketing spiels.

I like to think that it was because their talking heads at conferences and trade shows realized how idiotic they sounded when they touted their ILM capabilities, and simply refused to use the slide decks that argued the point. Whatever the reason, CA was now bringing the discussion of data management and ILM back within the realm of reason. I am all for it.

Going forward, I suggest, you need to manage data by type, first. Perhaps one day we will have the ability to manage data by class, as we can with mainframe ILM—but not today. For now, we must work with what we have.

Do whatever you can—policies, procedures, global namespaces—to try to reign in user files. Get specific tools to manage and archive e-mail, workflow content, and databases: they are getting to be a dime a dozen. With these tools you can effectively carve 40 to 80 percent of infrequently- or never-referenced data from your primary storage repositories, which will cause a noticeable performance improvement in your applications, enable more efficient backups, and maybe even forestall that expensive acquisition of storage you were planning for the next quarter. In short, you’ll be a star, I argue, stopping short of promising that women will dig you and guys will want to be you.

Heads are nodding in approval of so pragmatic and commonsensical a strategy. I seem to have satisfied the suspicious ones at the table that I am not here to endorse CA. All dig into their desserts and a few pass out their business cards to me and to their fellow attendees, promising to share information on how this-or-that product roll-out proceeds. I invite everyone to register in the Data Management Institute (http://www.datainstitute.org) to keep the discussion going.

That invitation goes to readers of this column as well. Write me with your ideas at jtoigo@toigopartners.comp>