NetApp Co-Founder Sees Expanded Enterprise Role

As more enterprise organizations move to distributed networks based on large numbers of Intel servers, Network Appliance executives believe the need for network-based storage is going to increase exponentially. Dave Hitz, co-founder and executive vice president of Network Appliance, explained why he thinks NetApp will become a more formidable force in enterprise storage--and the role he hopes channel partners will play in servicing those customers--in an interview with Editor In Chief Michael Vizard.

CRN: How is Network Appliance's channel strategy evolving?

HITZ: We are just, over the past couple of years, getting into the space where we were dealing with the type of customers that had heavy-duty service requirements. We're growing that part of our organization as fast as we can. And even there, we can't meet the needs of all of our customers.

Very consciously, we want to deploy our own people in professional services at the very leading edge of technologies like iSCSI and compliance issues. The goal we have is to package up the practices that we develop and hand them off to partners. It's not about altruism. It's about the fact that we're restructuring to more of an enterprise model, and the enterprise guys are saying, 'If you even want to do business with us, here's how much service we need.' And we can't grow it fast enough.

CRN: What unique competitive advantage does Network Appliance have?

Sponsored post

HITZ: NetApp has a single architecture. Except for our very lowest-end box, every single box we sell has the capability to do Fibre Channel SAN, iSCSI or NAS simultaneously. And the value of that goes beyond the capital preservation. It actually simplifies things like data provisioning.

CRN: What's driving NetApp's expansion into the enterprise, given its history of focusing on the technical markets such as life sciences or engineering?

HITZ: There are two domains of applications: technical and enterprise. All the oil companies have at least a thousand-node Linux farm. Every major tech app has already switched this way. In the technical space, this is not about when; this is not about how's it going to work. It has swept the industry. But if you're going to buy a thousand Linux computers, you're not going to turn around and plug Fibre Channel cards into them. If you buy a thousand Linux computers, buy a thousand Fibre Channel cards and then start adding up the port costs of the extra switches you need, the costs blow. If you're going to build that, the interconnect is going to be Ethernet-based.

That's the reason NetApp has been a big player in technical applications, because of our historical strength in Ethernet storage. It's also because SANs are very good for sharing a capital resource, but they're not very good for sharing data. If you look at clustered file systems, with 10 nodes, it's no problem for a SAN. Twenty nodes, it becomes interesting. A hundred nodes, it's a research project. We're going to be talking about environments with 10,000 nodes. We do that regularly. That's what NAS is good at.

Now here's what's cool: In the enterprise space, we're just getting started with this kind of architecture. For the enterprise side, the stuff we've already got now will make them happy for years. What they want is more uptime, more availability and to drive down costs while getting better services. CRN: How successful has your effort to get NetApp into the enterprise been so far?

HITZ: Before the tech crash, we hit $1 billion [in sales] and then dropped to $800 million. Now we've ramped back up to about a $1.5 billion run rate. Before the tech crash, 70 percent of our business was tech and Internet. Today, 70 percent of our business is not tech and Internet. The first time we were at $1 billion, the non-tech and Internet part of our business was $300 million. If you look at our business today at $1.5 billion, roughly $1 billion of it is that non-tech and Internet component. And that's during this flat market period. That is an incredible growth rate for the enterprise part of our business during a flat market period. It is all about simplicity and a single architecture that lets customers save money and drive down costs.

CRN: Do you see those opportunities extending into the midmarket?

HITZ: I think there are a lot of opportunities. If you treat medium-sized businesses not as little guys with wimpy requirements but as real, live public companies with a big set of as-yet-unmet requirements around SEC compliance, disaster recovery and 24x7 operations, this is an area for NetApp. Because of our single architecture, all of the systems we sell have those capabilities. They're sitting there waiting for someone to mine them out. It's one of those weird markets that you could either say there's no market there because people haven't been doing it or you could say what a great opportunity. It's a big unmet need.

CRN: From your vantage point, what are the trends driving these changes?

HITZ: Enterprise applications like SAP and Oracle moving to Linux, and there's this long-term trend of storage moving increasingly from direct to network. Plus, there's the whole issue of using inexpensive ATA drives for a whole variety of storage. And I think in the coming year, people will stop comparing iSCSI to Fibre Channel SAN.

CRN: Is network-attached storage supplanting the server processors in terms of relative importance?

HITZ: I think it has. Let me tell you how SAP thinks, because I was talking to one of their chief technologists. He said in the new model of SAP with NetWeaver, all these Linux systems and all these different apps can be instantiated on a different Linux system across multiple CPUs. The data for the apps sitting there on the storage is like the crystalline, hibernating version of the application. It's not associated with the computer. He said the app is sitting there crystallized on the disk, and when someone wants to access that data, we say this CPU is going to uncrystallize it and the app comes to life on that CPU for a while, as long as that data is needed. And then when it's not needed anymore, we recrystallize it.

I think that's the mental model of grid [computing]. You're still going to have lots and lots of CPUs and people are going to write the app to the CPU, but the question is going to be, where does the storage live and how do I get at that?

CRN: What's your take on the overall trend toward virtualization?

HITZ: I think virtualization is about easing transitions and about getting a common management interface over all your stuff. I see it more as a transitional kind of turf war than a real, long-term, sustainable kind of thing. Now the market could switch. All the back-end shelves could be commoditized, and everybody goes to virtualization front ends. That's a big part of the reason that we're shipping our Gateway product and playing in that space, because if the market shifts, we want to have our bet in the game, right? My personal guess is that's not how it's going to shift. But either way, we've got our iron in the fire.