He's in charge of merging two massive computing systems from Hewlett-Packard and Compaq Computer into a single infrastructure for the newly combined HP. Not only is the job essential to making the two companies one, but it's also a key part of whether HP can achieve the cost savings promised as one of the reasons for the $19 billion deal.

Even without the merger, managing the computer systems of a company with $81 billion in annual revenue is a round-the-clock job. Napier, 55, calls his BlackBerry pager his "CrackBerry" because he's addicted to it.

Although the merger was announced in September, the old HP and Compaq had to maintain separate computing systems for tasks such as accounting, human resources and online sales, for legal reasons. Napier had to overlay a new system that connected to the older ones and could be activated as soon as the merger was complete.

Outside the company, HP customers are watching as well to determine whether Napier's computer systems will prove that the company's products work--and thereby show that the purchase of multimillion-dollar equipment can actually save money in the long run. In all, Napier and his team have had to reconfigure more than 39,000 switches, servers and other networked devices.

Despite his high-pressure mission, Napier brings a "been there, done that" cheerfulness to his considerable responsibilities. He has been CIO at Compaq, Delphi Automotive Systems and Lockheed Information Management Services; weathered AT&T's spinoff of Lucent Technologies; and survived the global panic over the Y2K bug.

CNET News.com recently talked to Napier about post-merger life, overhyped computing ideas and his deepest fears, after his return from his first vacation since the deal began.

Q: Was the merger a CIO's dream or nightmare?

We had some day-one things we needed to get done. We needed a new HP.com that was open for business immediately, and it had to be able to sell both companies' product lines and combined bundled products, like the former HP's printer with a former Compaq PC. But a bigger deal was having the company employee portal ready on day one. We had 155,000 employees who had access to @hp (the company's internal Web site).

Did you have to buy any new equipment for the merger?
To some degree we did. We had to run two separate companies (as well as a new system that unified the old systems) over the course of eight months, through the ups and downs, as we went through the proxy fight. We had to have a separate environment that was isolated.

We did some clever things behind the scenes so we could interconnect the former HP company and the former Compaq with another layer on the top. One of the highlights was that we integrated a bunch of Active Directories so everybody's e-mail worked on day one. If somebody sent e-mail to bob.napier@compaq.com, it got converted.

What kind of hours did you work to handle the merger?
It hasn't really changed that much. For any CIO of any Fortune 50 or Global 100 company, it's a 24-7 job. It's one of the few positions in the corporation where you tend to see everything end to end. You need to understand finance and human resources. You need to understand (the) supply chain end to end, product development, engineering and marketing.

My wife was very proud of me last Thursday. I actually said I was on vacation and did not get on a particular conference call. She was extremely proud of that.

Is HP's computing-technology spending going up or down in the next 12 or 24 months?
(In 2002) we're going to take the sum of the former HP and the former Compaq and keep that flat...In 2003 and 2004, the budgets will come down. We've already committed numbers to the chief financial officer.

What percent will it come down?
About 10 percent in 2003 and (another) 10 percent in 2004.

We have now got pretty detailed plans going into the merger. If you look at the combined application portfolio, we're sitting at about 7,000. The opportunity will be to drive common processes and systems horizontally across the business and drastically reduce that application portfolio.

There were probably hundreds of legacy programs we looked at. We made some definitive decisions on what was going to stop, when it was going to stop, and when the resources were going to trail off, so we could use those resources to launch the must-do projects.

What's your biggest technology nightmare? What wakes you up in a cold sweat?
Being above the fold in The Wall Street Journal or The New York Times because I had some major failure. It's a break-glass, remove-resume time. It's end of career.

Here's my view of the world: I've been in this racket for 30-some-odd years. Back in the old mainframe days, the vendor built technology and you followed. Then I went through disruptive things called client-server. Then the Unix wars and distributed computing. I had an old professor who said, "If God had wanted us to have distributed computing, he would have put brains in our wrist."

Then there's this thing called the Internet. We could probably argue where we are on that curve, but we're in this era called "pervasive computing." Pervasive computing to me means it is an event when it's absent: If you pick up the telephone and there's not a dial tone, it's an event, because you always expect it to be there. Or go to the kitchen and turn on the faucet. Everybody expects it to be there. The absence of HP.com for 20 minutes is a major event.

What are the main areas you're excited about and will invest in?
Back in the late 1970s, we used to run multiple mainframes tightly coupled. We used to think that was pretty sophisticated. That's child's play. You look at a complex server farm today with 200 servers--Web servers on the front, application servers in the middle, database servers on the back end, and the way you wire those together to be one ecosystem--that's a really complex environment. We're working with HP's research and development organization and HP's software folks on coming up with solutions built into HP OpenView (software) that will help us better manage complex server farms with a lot more automation.

What's been your biggest disappointment in Web-based technologies? The Internet has been full of revolutionary ideas.

I had the same disappointment around Web technology that I had around ERP (enterprise resource planning software, for accounting and other business tasks). It was oversold, overhyped and before its time. It took awhile for the products to mature to where you could really get the benefit.

The technology has never been the problem. You look at all the companies that had major ERP failures, and it's because the capabilities of the software were overhyped and oversold, and the company had a hard time getting process changes in their own environment (to match how the software works). We got through that era. Now having ERP is the price you pay to stay even. People are a lot smarter about how to implement ERP systems.

What are three technologies you think will have a big effect in the next year or two?
I'm a big believer in Web services. I recently said to an audience: If you never got the ERP thing right, what you may want to do is go back and look at...using an architectural strategy around Web services. If you're a big company with business in 100 countries, then going to a UDDI (Universal Description, Discovery and Integration) directory, grabbing an address (and) then going to the one place where a product gets priced creates all kinds of greatness. The one configurator that configures a particular product line gets rid of a whole bunch of system redundancies.

I think wireless will continue to grow. The access devices (for connecting to the Internet) will continue to get more sophisticated and more powerful. Whether you're Merrill Lynch, HP or U.S. Steel, at some point you're dealing with heavy digital content you're delivering to the edge of the network, anywhere, anytime. (But) I think we've got a lot further to go in wireless.

And in the bigger organizations, real-time access to customer data is a big deal. CIOs have been trying to solve this for 30 years. We've still not gotten to where there's one single, refined repository of customer data where you can go and profile your customer. You end up with these islands of data. I'm looking to solve that problem in HP using ZLE so I can turn around and show it to our customer base. (ZLE is Compaq's Zero Latency Enterprise strategy to link disparate computing systems with one mammoth, high-speed database.)

CIOs typically have a preference for centralized computing power--servers and mainframes, not PCs. Is it a good idea to shift computing brains to PCs?
It depends on what problem you're trying to solve. We (at HP) have a huge global footprint of PCs. I don't consider them a problem to manage, because we've gotten very smart about how we manage them. Whether I'm logging on from here in Houston or from a satellite office in France, a script runs my logon, looks at my system, validates that I'm running a valid image, validates that my virus protections are up-to-date.

So you've managed to achieve centralized control over distributed resources?
Yes. Occasionally the computer will say, "I have to update this. Do it now or wait?"

How long should a PC last? Isn't it expensive and difficult to replace those systems so frequently?
We're on constant technology refresh. On average, we're shifting things every three years.

The CIO in me says I would like them to last forever. I don't get them for free. I take a budget hit. You'd like them to last forever, but technology moves on. I think three years is probably right. Would I like five years? Sure. But I don't see it as the biggest problem I have.

Do you believe outsourcing computing operations is a good idea? In particular, would you let another company handle very important parts of your business?

What makes sense to CIOs are very pointed solutions. If somebody is an expert at e-mail and you're not, and somebody can do it better for you, go for it. If somebody has got an ASP (application service provider) handling something like dealing with expense reports and provides that as a Web service soup-to-nuts, and it's not your core competency, you go for it.

I might consider allowing somebody to fly the plane for me, but I'm going to be the one who owns the flight plan.

What needs to be done for Windows servers to step into the roles most Unix servers are handling today?
I would say the toolsets we use to manage Unix are more mature. Managing a complex Windows-based server environment, you would like to have the set of tools you have in the Unix world.

How do you measure success or failure as a CIO?
You can go through benchmarking, but it's hard to find businesses that mirror your own. I rate success or failure on how well I think the organization's strategy and governance works. How do you keep 10,000 IT professionals all on the same page? If you think about running a very broad and deep global IT organization, the fundamental thing is you have everybody speak the same language. I don't mean French or German or Chinese or Japanese--I mean the same language of business.

Back to intro