Just prior to announcing that his company would miss its quarterly earnings mark (an announcement that drove the company's stock down by 35 percent in one day), Veritas CEO Gary Bloom chatted with ZDNet executive editor David Berlind.

Bloom waxed philosophical about the definition of utility computing, how CIOs should make a utility model their priority, and why, with virtually no hardware or application software agenda, Veritas should be perceived as the unbiased go-to company once the decision to utility-orientate a computing infrastructure has been made.

Bloom claims that, contrary to what Nicholas Carr believes, IT does indeed matter. But, to Carr's point, he sees the rising cost of infrastructure as a monster that's depriving businesses of the investment capital they need to develop the sort of applications that can make a competitive difference.

ZDNet: If, instead of being the CEO of Veritas, you were the CIO at some enterprise and you had a meeting with your CEO this morning and he or she was waiting for a proposed strategic IT plan from you, what would you be proposing?

Bloom: If you look at the article in Harvard Business Review by Nicholas Carr, he has a premise that IT doesn't matter and that IT has lost its relevance. In many respects, from the CEO's point of view, that's probably true. There's been an ever increasing cost to the fundamental infrastructure of IT. The cost of labor and the systems to run the applications is going up. The fundamentals of just having computing have gotten very expensive. Companies have a backlog of applications that need to be developed, deployed, implemented, and rolled out to their employees and customers.

The main premise of my proposal would be that we have to find a way to dramatically lower the cost of the infrastructure so that we can put our investment dollars into IT and into the business applications that help differentiate the business. The only way we're going to do that is by making the infrastructure dramatically more efficient, which means making better use of the hardware, better use of the labor to run it, and coming up with a more efficient operational model for the IT infrastructure. If we do that, then we'll have more investment dollars to build better business applications. Those business applications can drive customer intimacy. They can drive better partner to partner communications. They can drive more efficient information about our marketing, about our products, about our inventory, and about our build times. These are all the things that matter, depending on the nature of your industry.

ZDNet: So, if you're in front of the CEO and telling him that your goal is to reduce infrastructure costs so as to free up investment dollars for differentiating applications, what, architecturally, would you be proposing?

Bloom: Fundamentally, it would be to implement a utility computing model. I define that model as nothing more than systems that perform very well, having systems that deliver a very high level of availability, and that run on a shared, automated infrastructure for low cost. It's no different than a water utility. You turn on the tap. It's available. It comes out with water pressure (that's performance). You don't dig a well in your backyard to get water. You use a shared automated infrastructure and it's pretty cheap. There's no reason why IT, internally within a shop (I'm not talking about a public utility), cannot run on a utility-like model. So, I would propose that we migrate the infrastructure towards a model of utility computing that would free up the investment dollars for application engineering and development.

ZDNet: But, I don't have to run the utility myself, do I? I could go to somebody like a Salesforce.com?

Bloom: Well, you could certainly look at it either way. One way is that you can outsource it to a company like Salesforce.com, where it's more like a public utility. The approach is that you could run a similar model internally, depending on the size of your operation with a focus on the same benefits: performance, availability, and a shared automated infrastructure. Whether you provide that and take the cost benefits, or you tap into a public utility and benefit from the cost benefits it has to offer is relevant based on size of your company more than anything else. From a Veritas point of view, we're indifferent. We can provide a customer with the technology to provide their own utility. On the other side, we can provide a public utility provider with the infrastructure so that they can build and support a utility for their customers. It's really a model that, from our business perspective, works either way.

ZDNet: Now that you've brought up Veritas, where does Veritas fit into a plan to build an IT utility?

Bloom: There are two ways you can look at it. One is to look at the markets we're in. Two is to look at the building blocks we have versus the rest market in order to enable somebody to build a utility computing model. If you look at it from the point of view of what is a utility today, and how do people typically get started and what is it that we have that's utility-like today, I would argue that storage is probably the closest thing that IT shops have to a utility today and it's the place that we recommend to our customers that they start with for utility computing. We have a contrarian view of utility computing at Veritas. We don't have a hardware agenda. So, it all becomes a question of how do I get started, and we think storage is a pretty good place.

In most IT shops today, storage is very close to the utility model. Why? Look at how a utility is defined. It typically performs pretty well, it's typically very available, and it's typically very centrally managed, centrally purchased, and centrally supported by the IT group--and we're not just talking about disk drives. This extends into things like backup and recovery. In a typical enterprise account, or even a small business account, you don't see all the employees leaving on Friday afternoon with a copy of their data under their arm in a backup tape or on a disk that goes in the trunk of their cars for off-site storage. That's typically provided as a service by the IT group. Just the same in most IT organizations, the various business leaders around the IT operation aren't allowed to go out and buy their favorite disk drive of the day. Those are typically centrally purchased by the CIO. And then that CIO and his operation typically take responsibility for the performance, availability and automation of that technology. And that's very much the characteristics of a utility. It performs well. It's highly available. It runs on a shared automated infrastructure that's very centrally managed.

We go [our customers] and say, "Look, in your storage setup, you already have the initial building blocks of the utility model. Now what are the other utility components or building blocks that you need?" Certainly, if you drive up the availability curve, then you need server availability and application availability. Those are addressed by Veritas' clustering software. We've started hooking that into the applications.

In the performance category, it's more than just the performance of storage, if you want a complete IT utility. It has to be the performance of servers and applications as well. Our Precise Software Solutions Ltd. acquisition positions us with application performance management technology to do the end-to-end performance management. In the automation/shared infrastructure category, we've led that market in the storage software space in both our foundation products (such as file systems and volume managers), as well as in our backup technologies.

Through acquisition we've added the building blocks to do the same level of automation for servers and applications. Those acquisitions were the Jareva acquisition and then the acquisition of a company called Ejasent for application provisioning.

ZDNet: In describing the attributes of a utility, you've identified a standard threshold for what's acceptable in the storage market. We know that things are moving towards standards on the hardware side, especially around the Intel platform. In the IT food chain, does the point at which the most value can now be delivered live somewhere in between storage and servers. Is that your sweet spot?

Bloom: I don't think you can look at it and say there's a layer between storage and servers unless it's all delivering high levels of availability and performance and it's all running on a shared infrastructure. You get incremental benefits for each piece that you automate. Is there value in setting up the storage utility in a typical IT enterprise? Of course. Is there more value if you set up the entire IT operation as a utility? Well, the savings will be that much bigger.

ZDNet: Right, but in both respects, the piece that delivers the reliability and the performance of the utility actually sits in a layer above the storage itself, right? Therefore it's in between.

Bloom: Absolutely. What we call storage software actually sits on the server. That's the differentiator of our solution. When we say we don't have a hardware agenda, we don't manage storage array by array. We manage storage at a layer level. This is particularly important in talking about the trend towards Intel-based computing. The reality is that the trend that's really going to come into play and drive utility computing long term will be what's called blade computing. Not only will there be low cost storage blades that store a phenomenal amount of capacity and deliver very high levels of performance, there will be server blades as well. If you don't move towards a dramatic level of automation for blade computing, your operational costs will offset the cost savings on processors and storage capacity.

ZDNet: But wait, don't blades represent your biggest threat? Utility computing and automation is all blade vendors can talk about. When they sell you blades, they want you to buy their automation and utility software too. If I do, then why would I need Veritas?

Bloom: Certainly the hardware vendors would love you to believe that all that software is purchased for their particular platform. What that fails to recognize --- and the hardware companies suffer from this --- is that the world is heterogeneous. If I was a single-vendor IT shop and I had nothing but IBM equipment and IBM databases and IBM application servers, then I'd expect IBM to provide my entire infrastructure to make that efficient and operate at a low cost. If I'm an Oracle customer and 100 percent of my data is in a Oracle database and I use the Oracle application servers and I use the Internet file system since there's no data outside of the database itself and the machine is dedicated to Oracle, then I'd use 10g.

ZDNet: But you brought up blades. The thing about blades is that there's no standard in the blade market. That pushes customers towards a single hardware vendor. People who have adopted blades have largely said "I have to go with a one vendor solution because none of them work with each other." In that scenario, wouldn't customers just assume go with the blade vendor's software?

Bloom: You're right. There are no standards. Just as there are no standards in most Unix systems. What HP Unix does compared to Sun Unix in comparison to IBM Unix compared to Linux are all dramatically different. What Veritas has done is that we've come in with layers of technology that make all those environments look common. So the way we cluster systems for high availability is the same for Sun, IBM, HP, Linux, and Windows. The customers can train their employees once and operate their systems across different platforms consistently regardless of hardware manufacturer. The very fact that there are no standards in blades is why there needs to be a layer of technology from a company like Veritas. We rationalize the fact that there are no standards yet by providing a common way of managing all systems. If we don't give customers a common way to manage systems, then they have to train their people for every piece of hardware they buy. Customers have routinely indicated that they don't want to do that.

That's precisely why -- if I use [this clustering] market that's been around for a longer period of time -- Veritas has done so well there. We've outsold Sun on their own platform even though we've competed with them for five to six years now in the clustering market. If you talk to Sun, their premise would be "Why would anybody buy clustering software from anybody but their systems provider?" The reason is that the customer doesn't just have Sun machines. They also have IBM machines and HP machines and they want a single way of managing all of them. The reality is that blades are nothing more than a continuation of a heterogeneous world. A heterogeneous world is one that says there's a big opportunity for the software that comes in without a hardware agenda and simplifies it across all of these hardware implementations. That's what we do for a living and why we've been able to generate real solid results for a number of quarters, even in a downturn.

ZDNet: As long as there are no standards, and companies don't look to consolidate on one vendor, you're in a great spot. But if a corporation, to gain certain efficiencies, says "we're going to set a standard for all hardware" or if a standard comes out that levels the playing field, how does Veritas respond to that?

It's funny. We actually like standards. We think we can differentiate based on the capability and level of automation our technology can bring to the customer. The primary thing about standards is that they've helped us reduce our cost of engineering. We can take advantage of the fact that there are standard interfaces for managing hardware. The reality is that when you put all of the hardware vendors in the same sandbox to play together, they rarely play very nice. They typically throw sand in each others' faces. So, we as an industry never get any real standards.

ZDNet: That's true of the blade market. Everybody wants to win on the basis of their on-the-fly provisioning. They also want to be the provider.

Bloom: IBM will be thrilled to manage and provision any IBM blades. Sun will be very happy to manage Solaris blades. HP is happy to manage HP blades. The reality is that most shops aren't going to have just IBM blades, or Sun blades, or HP blades. They're heterogeneous. If you're the CIO, even if you wanted to be 100 percent homogeneous to a single hardware vendor, the chances of you successfully doing that are slim unless your company is doing absolutely no merger or acquisition activity.

I had a financial services customer who said "I don't need your solutions because I'm 100 percent EMC. I buy all my software to maintain my disks and do all my storage provisioning and storage management from EMC and that's all I need." I asked him what he'd do if he did an acquisition. He said he wasn't doing any [acquisition s] that were big enough where he couldn't replace whatever got brought in with EMC, thus staying homogeneous. Less than 60 days later, he called me back to say he was doing a merger of equals -- it was one of those multi-billion dollar financial services mergers -- and the shop he was merging with was 100 percent Hitachi. Even if he wanted to swap it all out, he knew that on the day of the merger, he'd have to be able to communicate between his EMC arrays and his Hitachi arrays. The equation changed on him, even though he had a desire to be "single-vendor."

ZDNet: What about the application layer? You provide replication. You provide clustering. Sometimes, in order to get the benefits from a particular software provider's solution, you have to use their replication or their clustering abilities. Examples are Sybase with their database replication capabilities to facilitate mobility and Oracle with its database clusters. From the customer's perspective, are you asking customers to bypass those vendor-specific solutions and accept your way of doing it? Couldn't that interfere with the functionality of some applications?

Bloom: You're into two different markets there. There's application-level replication, which is for the purpose you just described, and then there's what I would call storage-based replication to drive the ongoing availability. We're moving blocks of information at the disk level from one location to another; so, if you want to duplicate a complete environment for disaster recovery or you want to move large volumes of information, you would move whole disk drives worth of information from location A to location B. This is typically done over a wide area network and typically from one disk array vendor's technology to another. That's when you 'd use Veritas. When you get into moving employee data from the US to the UK, but just for UK-based employees, that's at the application level. That's not what we do.

ZDNet: So the same rules apply to database clustering as well? You're suggesting that people stay with Oracle's clustering technology and then you provide the underlying movement of the entire cluster from location A to location B for the purposes of disaster recovery?

Bloom: The big difference is that we cluster the entire environment. Oracle is just clustering the database. So, as long as your failure is purely in the database and the machine is dedicated to Oracle, you're OK [with just Oracle's clustering]. The reality is that most systems have other data in the file system and you may want to cluster the application server as well. You want to cluster the entire environment for failover. That's what Veritas does.

A good example is what EBay does with our technology. They use our technology to deliver availability irrespective of what component in their stack could be failing at any given time. So they can deal with an application failure, a database failure, a server failure, a processor failure, or a storage failure. For all of the different reasons why a system might stop running, we provide the failover technology so that the customer doesn't see degradation in service.

ZDNet: Who uses Veritas' solutions? Bloom: Microsoft leads the clustering market per se because, by virtue of the technology included in Small Business Server, they took an allocation of the clustering market. Up until they started doing that, we were leading the market. But if you look across the Sun, HP, and IBM environments, we're the clear leader in the clustering marketplace. We sell to 99 percent of the Fortune 1000 companies. The vast majority of large corporations rely, somewhere in their operation, on Veritas clustering today.

ZDNet: Well, if you're in 99 percent of the Fortune 1000, given the way that each of the major vendors tell me that they want to be the one-stop shop for utility computing, especially to that crowd, it would seem to me that Veritas can't stand alone much longer. Sooner or later, it will have to merge or be acquired by a much larger organization.

Bloom: Well, I'll never say never relative to the potential of Veritas being acquired. But if you look at our size from a revenue and market capitalization perspective, the number of software companies that can acquire us is relatively small. You're talking Microsoft, SAP, and Oracle.

ZDNet: And of course, Larry Ellison's shopping list was just recently published. Yes, and we weren't on it. So those are the three companies on the software side. If you look at it from the hardware side, there are a number of hardware vendors who are much bigger than we are. But, one of our major differentiators in the market is that we don't have a hardware agenda. So, we're not the ideal acquisition target for most hardware companies.

ZDNet: Well, if for no other reason, your customer list.

Bloom: That's why I say, "Never say never." I never say never about what we might acquire and how far we'd go, and I never say never about who could come after us either.

We're also a company with close to $2.5 billion in cash; not a bad thing for a company that has about $11 billion to $12 billion in market cap. That sort of cash is not a good thing when it comes to being a target of acquisition. From a strategy perspective, there's nothing that says that Veritas can't lead the utility market on a standalone basis. The reason being is that everybody else going after the utility market has some bigger agenda. If you're Oracle, your agenda is all data and all information in Oracle and 10g isn't going to do anything for the customer that also has SQL Server and DB2. If you're IBM, you're not going to do anything that makes your customers more successful at running Sun processors. If you're HP, you're not going to do anything to help IBM to sell more IBM processors. If you're EMC, you're not going to do anything that helps IBM sell more Shark disk arrays.

The fact is, if you step back from all of that, customers run a mixture of all of those things. I don't think the typical customer wants an Oracle utility for their Oracle environment and an IBM for their DB2 environment, and a Microsoft utility for their SQL Server environment. Nor do I think on the hardware side of things that customers want an EMC utility to run their EMC environment and a Hitachi utility to run their Hitachi environment. I think what customers want is a common utility model that works in a heterogeneous world. They want to manage their systems from an availability, performance, and automation perspective across the storage, server, and applications layers in a consistent fashion regardless of what vendor they choose. This way they can run a single utility instead of multiple utilities.

ZDNet: How has regulation changed your business?

Bloom: Generally, it's been helpful, given that a lot of our revenue is in our data protection and storage software businesses. Regulations have suggested that you have to save all of your data for a much longer period of time. In some sectors, it is literally all of your data. That causes the volumes of storage under management to increase dramatically, which means that software to help automate it is a good thing to invest in. If you're not growing the amount of storage you have, you may not have to automate it as much as if you were growing it. Certainly, in the data protection business, the combination of the events on 9/11, the ongoing focus into disaster recovery, and now the whole category of regulatory compliance are all contributors to growth.

ZDNet: With all that data out there, how does one find what they're looking for?

Bloom: That's the big challenge that the industry is trying to solve that hasn't been completely solved by anybody. That's kind of a green field opportunity in the software sector. Most CIOs will tell you that "I have all the data I need. I just have trouble finding the information within it." That's why there's data lifecycle management technology coming from Veritas and others that talks about how do you manage the information and not just the data.

ZDNet: This has been a vexing problem. Unstructured data lives everywhere. It lives on servers and people's hard drives on notebook computers. It's very distributed. It seems as though if utility computing is one of those areas that vendors are clumping around, then this idea of mining data from an unstructured cloud is the other. Microsoft, with its WinFS technology, is trying to figure out how to strap a metadata layer around all that unstructured data. Where is Veritas coming at it from?

Bloom: When you get into the information, customers may tell you that they know they have the information and therefore they are compliant [with the regulations]. But it's a real business decision to want to be able find it. It's not clear that everybody wants that.

ZDNet: But, from a content management perspective, aren't there are people that want better management of the unstructured data.

Bloom: The content side of it is really an application that we have not moved into. I would argue that it's more a market for the ERP companies or the application companies than it is for the infrastructure companies. A good example of that, going back to Oracle's shopping list, is Documentum. Documentum was on Oracle's shopping list, but then it got bought by EMC. When I look at Documentum, all I see is an ERP application owned by a storage company. I don't see it as a logical extension of storage.

ZDNet: If content lives in a variety of different places, just the same way that Microsoft's WinFS will supposedly one day provide this data DNA that crosses all Windows desktops and servers, to give users visibility into all data across all platforms in their enterprises, one would need a similarly common DNA, except for the fact that it wouldn't be Windows only. As you just described Veritas, there are few Switzerland-like companies in the world that are positioned to do something like this other than Veritas. You even have your own unified file system which is a prerequisite for such an idea. You really don't see it as a natural extension of your business?

Bloom: Well, you're right relative to providing uniform access to data. We provide uniform access to data. One of our new developments that we started rolling out around our volume management technology is something called portable data containers. You can actually take a disk drive from one manufacturer's computer and plug it into another's. So when you talk about moving data or information from a Sun machine to an IBM machine, it's as simple as unplugging a wire from one and plugging it into another. You don't have to reformat the disk for the other operating system. So we're doing that. But where I take exception to your point is that in managing content, there's no difference in managing documents from an ERP perspective than the way a financial application manages a general ledger. Our technology runs under a dramatic number of general ledgers. The vast majority of ledgers and HR applications in the world run on Veritas solutions.

ZDNet: But general ledgers are transaction based. Every transaction is going to be stored as a record in a database. The same cannot be said for unstructured content and data. It could be living in a word document on a hard drive somewhere. That's not a database. Maybe a form in an insurance application would live in a database. To the extent that you have a file system and metadata is a natural outgrowth of a file system, it just seems that content management is a natural extension of your business.

Bloom: I think there's some synergy when you're talking about metadata layers. I guess where we're on parallel planes but not intersecting is in the discussion of the actual physical applications at the application level that manages documents. There are multiple vendors in that space: Documentum, Filenet, and others. All of those applications ultimately need to be able to find both structured and unstructured data, but they're separated from the infrastructure.

ZDNet: But shouldn't that application move into the infrastructure? You're the one who said that we have to architect an automated infrastructure so that companies can spend more of their resources on applications that make a difference. This is an infrastructure application isn't it? Businesses should be investing in something else for differentiation.

Bloom: I don't know that I agree with that. We could attempt to redefine the market and start to say that everything is infrastructure and that our addressable market is $50 billion instead of $8 billion. That's not how I view the marketplace. Do I think there's a layer of technology that provides uniform access to information, common ways to manage the systems and infrastructure of IT, and help drive down the costs within the defined characteristics of a infrastructure layer? Yes. That's what Veritas is doing. As soon as I start bridging up and saying "Now we're going to provide the application to manage the data," then I might as well go become an ERP provider. We could buy PeopleSoft, and I'd be the white knight.

ZDNet: I hear they're not for sale.

Bloom: It's a good discussion. You're taking the concept of what we're talking about to the next level in an interesting way. But when we look at our strategy at Veritas, we have a pretty practical view through the eyes of the CIO. The CIO today is simply struggling with the question of "how do I rationalize all of this complexity and this heterogeneous technology so that I can drive down the cost of my infrastructure" while also starting more discussions around topics like the ones you brought up --how to get better search capabilities and how to get better mobility. These are all the right questions. But the reality is that there's a huge business opportunity to address and we call it utility computing. It's simply solving that fundamental infrastructure problem to free up intellectual cycles and the financials to go invest in these bigger more important topics.

ZDNet: So, IT matters?

Bloom: I think IT makes a huge difference in the world. I think it's also true that money has been going to the wrong place.

You can write to me at david.berlind@cnet.com. If you're looking for my commentaries on other IT topics, check the archives.