IBM`s Big Storage Picture

IBM storage group has done very well with SVC--SAN Volume Controller, IBM's key virtualization component--but is it enough? How do you see IBM differentiating itself here in the future? What areas do you see as growth areas for IBM's own R&D investment?

First of all, SVC has been an extremely successful offering for us, not only for its intended benefits but for its unintended benefits. I put part of the business case together to do SVC. We never really understood the capabilities of what clients were really looking for. What we knew was that this would be the next evolution of SANs.

The idea of SAN was to be able to better share, and that this would be the next evolution of presenting a large pool [of storage capacity] to the application, allowing for more flexibility. That's worked just great.

But here's the unintended benefits: In a hypergrowth market--and I think this is the trend that is beginning to be understood, and let's just agree that IT budgets are relatively flat--[we're in] a world where people not only have to store everything for what I'll call fairly ambiguous compliance regulations, but they want to store everything because of the business intelligence and mining capabilities of the data.

I was with a Midwest bank that used to drop all of the records on their credit card business that weren't required for the transaction to be recorded and brought back to the client. Now they save all of it, so that they can deliver this as information for tailored marketing. [Banks are] all looking for new business.

This whole notion of driving new business off the data is just emerging. You see it also in the telcos. The [IT] folks who used to just be responsible for driving costs and efficiency now have to be flexible in everything else.

Why is this important to SVC? Part of what you want to do in hyper-growth is get better utilization of the assets you have. Most storage assets, except those connected to the mainframe, run at the same kind of utilizations that Intel servers do: 10 to 15 percent. Deployment to virtualization allows for over-provisioning. It allows you to get about double the capacity for utilization.

Step 1: Deploy for virtualization, and your next set of growth is paid for. Step 2: You can now repurpose assets. No one in the industry can copy data cleanly from one tier to the next, unless it's the same asset.

Is there software available to make those connections?

There is direct connecting software, but it doesn't give you the overall flexibility to roll in and roll out whatever you have. Technically, the differences are significant, and they’re fairly complicated. What the power of SVC is: I put it in my infrastructure, I acquire a small company, and they have assets; I can roll the assets right in, and they become part of the ability to repurpose things that I already have.

[SVC] is my highest referenceable offering in the marketplace. I've got hundreds of customer references on SVC. We've got about 4,000 deployments now; so we get about five new customers every day.

The other unintended benefit is this: When you are a client and you have only one vendor … it's really hard to switch. Before storage software became in vogue and was really deployed, back in the '90s up to 2002 or so, you used to hear in our business, 'Sweep the floors--all of his stuff out, all of my stuff in.’  You never hear this anymore, because the stickiness of the software is so great.

What SVC does is it puts an abstraction layer between the application and the physical storage devices, and that actually shifts the point of control from the vendor back to the client. That's a benefit we really never understood, but it has been extremely attractive as entree to competitive accounts.