by Kevin Fogarty

Got a Good Grip on Virtualization? You’re Just Getting Started

May 15, 20085 mins

After the whole company floats on an ocean of virtualization, users are still going to complain if they can't do the things that they want to with the data that you've made it so simple to access.

You know, I hate to tell you this, right when this virtualization thing was going so well for technology users, but whatever virtualization projects you’re working on right now won’t end when you finish building out the storage area network (SAN) and VM server farm. Not by a long shot.

Sure, you already know there will be follow-on projects. You’ll have to add some level of desktop virtualization, probably for departments that have pretty low demands on IT anyway (and probably not much pull with upper management, either, but that’s another issue). You’ll have to virtualize some apps, beef up your network backbone to handle the additional traffic, and upgrade your systems and network management to keep your VMs and PMs (physical machines) on track.

You’ll also want to look at data-leak prevention, VM-specific access control, backup and storage management; VM-enabled business continuity, maybe.

If you’re optimistic you might spec out the full virtualization picture –in which the network, file system, storage and applications are all virtualized and controlled by a kind of IT-manager-as deus ex machina, shifting resources and capacity around behind the scenes while end users work blithely on without knowing where any single one of the bits they manipulate lives or why.

Undoubted optimist, IT storage analyst and former integrator George Crump who— aside from an unfortunate tendency to capitalize IT End User and view storage as the axis on which the IT world revolves—sees that universal virtualization as a logical, if challenging, final goal of major virtualization projects.

But after all that work—flooding the IT infrastructure with abstraction layers until the whole company floats on an ocean of virtualization, high above its data and computing gear—users are still going to complain they can’t do things with the data that you’ve made it so simple to access.

Virtualization projects are IT architecture improvements, but good architecture doesn’t always make for good buildings.

Uber-architect Frank Lloyd Wright, for example, is famous for building beautiful houses scaled to his own diminutive height, so that inhabitants taller than 5′ 9″ either walk around stooped, or with door-lintel-shaped lumps in their heads.

That’s roughly the position you’d be putting end users in if you restructure your infrastructure, but leave the data as is. End users would be able to get at data so easily, they’ll immediately wonder why they can’t do more with it.

It’s the most basic problem in IT, and the most dull.

Forget virtualization. Data formats, database structures, data restrictions, incompatible applications and quilts of middleware are the real IT infrastructure of any company.

All the rest—no matter how incomparably more cool—are just access mechanisms.

Six or eight years ago the most successful CIOs I talked to were obsessed with access and interactivity. The key to success was external—finding ways to exchange data between their company and their partners,’ even with lowest-common-denominator data formatting and network connections that were more like rope bridges that superhighways.

Now the most successful CIOs talk about data structure as the absolute bottom layer of their infrastructure—the foundation on which anything else can be built.

With the right formats — meaning the most vanilla, translatable, accessible data formats and structures—they can do absolutely anything with data, without having to put it through contortions and conversions that slow the whole process down and introduce interesting new ways to break applications.

Data-format projects are even less sexy than most infrastructure projects. No C-something executive is going to drop by your office with an airline magazine to talk about data cleansing, de-duping, format conversion or migration.

The marketers and C-somethings probably won’t even be able to articulate what the problem is. They’ll just know “the systems” won’t let them do what they think the need to do.

“The system” can’t handle the mishmash of legacy data in a way that’s quick and consistent enough to let them put together new promotions or market-analyses or customer-retention programs or premium services.

But if you’re trying to make the new building work with the old, ratty, rotted foundation built by some ancestral CIO with too much to do and not enough time to plan for the future—your end users will dump all over your shiny new virtualized infrastructure and wonder why they spent all that money just so they can not do faster the same things they could not do before.

Sorry to break the bad news. The weather’s warming up, vacation and movie-blockbuster season are coming, and the big summer entertainment was going to be watching VMware and Microsoft whale on each other while the hardware vendors and open-sourcers pitch in on both sides.

That’ll still be fun.

But you’ll have to expand your own ideas about restructuring your infrastructure and rearchitecting your architecture (eventually you’ll have to stop using those terms as verbs, too), and wander down into the pocket-protector zone to talk data formats with your DBAs.

It’s overwhelming; it’s boring; it won’t get you any immediate pats on the head from your bosses. But I guarantee the data trolls will be thrilled to hear someone understands the value of data structures, and will have some good ideas about how to fix them.

And if you get that effort moving while you’re still working on your virtualization stuff, you can probably roll the whole thing into the same set of project budgets. That’ll make the whole thing a little less painful.


Veteran journalist Kevin Fogarty has covered the technology and business worlds for more than 18 years. Fogarty is a former editor or analyst at Computerworld, Baseline, eWeek, and Illuminata. Fogarty won an ASBPE (American Society of Business Publication Editors) award for column-writing during his stint at Computerworld.