Ellison was right. Oracle would have been in a jam if Lane had suddenly quit in early 1997. But there was a substantial price to be paid by both men for Ellison’s bribery. If Lane was getting inflated ideas of his talents thanks to the flattering press coverage that Ellison was happy for him to receive, learning that Ellison thought him so important to Oracle that he was willing to make him fabulously wealthy was bound to puff his ego up still further.
Don Lucas says he warned Larry about the possible effects. “I’ll never forget it. I very, very seldom have disagreed with Larry. . . . This was a huge stock option, like a bribe as opposed to an incentive. That’s going to twist people’s minds. It has all these bad effects. Ray says to himself, ‘I’m worth all this money, I’m irreplaceable, I’ve got a lot of power, I’m like Superman.’ So it reinforces all the bad things. If we’d given him regular incentives and if Ray would have stayed, I think that would have been better for everybody. If he wanted to be loyal to the company and get compensation, reasonable compensation, he stays. If he wants to go be a big shot, be a CEO, have all that power, that’s fine, leave. It would have been better for everybody. So I think that was a critical juncture. I’m not saying I was right and Larry was wrong,4 but I think you have to stand by your principles and I think to overcompensate can have deleterious effects.”
From his perspective as a senior member of Lane’s team, Jay Nussbaum was convinced that something had changed. “The Novell thing pissed off everybody at Oracle. Ray would always argue about loyalty, trust, and integrity. But everyone saw what he did. He put on a mask and took a gun to Larry’s head. . . . It gave a signal to the world that Ray didn’t respect Larry and he thought he should be CEO, he had this blind ambition. It became obvious that there was no love affair. So much so that Larry starts to wonder what life would be like without Ray.”
Even though Ellison used the threat of Lane’s departure to get another 1.5 million options for himself and 1 million for Jeff Henley, he felt increasingly sick about what he’d done. I don’t think he really blames Lane for what had happened, but he thought that he had in some sense been manipulated, and the feeling festered. Carolyn Balkenhol, who took over from Jenny Overstreet as Ellison’s close assistant in 1995, has an interesting perspective: “I think the things that always upset him most are his own issues. I think that when he seems really angry it’s because he’s done something that he perceived to be a mistake, particularly when he’s made a decision that he thinks has set in motion a series of events which didn’t have the desired result. I think most often if he’s really angry it’s because . . . he did something that took a wrong turn.”
Ellison began to realize that not for the first time, in his eagerness to carve out just the portion of the CEO’s job that he enjoyed, he had delegated too much power and responsibility. It’s a characteristic of Ellison that when he thinks he has found someone who is competent and who can help him, he invests an excessive amount of faith in his or her ability. Just as when Gary Kennedy said he needed more authority to get the job done, Ellison would give it to him without much thought, the same applied to Lane, albeit on a much grander scale. Ellison says, “I’ve been to this movie. I’m not going to let this happen again.” At some point, in Ellison’s eyes, Ray Lane stopped being the solution and started to become part of the problem.
Within a year of Lane’s receiving his option grant of January 20, 1997, Ellison had begun the process of peeling away his direct reports. Strongly influenced by Gary Bloom’s critique of Lane’s management and his plea that Ellison should assume more of the responsibilities that normally go with being a CEO, Ellison began to shift whole departments from Lane to Bloom. Among the first were marketing and education, but inevitably, the first battleground was applications. Ellison’s realization that the commitments that Oracle was entering into with customers such as Kellogg and the never-ending litany of lengthy and money-losing implementations, especially those involving the CPG best-of-breed suite, had convinced him that development must regain control of product engineering.
By this time even Lane was prepared to admit that best-of-breed had proved a great deal harder and more costly to execute than he had initially realized, but tensions increased between him and Ellison over the broad strategic direction that the applications business should take. Although Ellison had put a stake in the ground in Paris two years earlier, to mark his profound disillusionment about the unmanageable complexity that was inherent to client/server, Oracle was only now near the point of releasing software designed for the Internet. Once Oracle became committed to engineering client/server software, it was inescapably locked into that architecture for years, even if theoretically a better alternative might have existed. Ellison’s public statements about the coming of Internet computing were not just about attracting media coverage and tantalizing customers, they were also a deliberate tactic to open up the debate within Oracle.
Ellison says, “You can’t tell engineers, ‘Just do it.’ You have to persuade them that the Internet is a better architecture than client/server. You have to persuade them the Internet is going to win in the marketplace—even though Microsoft is backing client/server all the way. They have to believe in what they are doing. If you simply bark out orders, you get malicious obedience—they’ll build exactly what you want and make sure it doesn’t work. I had to proselytize in every meeting. It took years before we killed off our last client/server product. It wasn’t fun. I’d say, ‘You know that thing you’ve been working on for the last year and a half? Throw it all away. We don’t want it.’ It pissed people off. They’d say, ‘I’ve been working hard for a year and a half, and now you want me to throw it all away?’ ‘Yeah, I do.’ ‘Are you crazy? I quit.’ It was very difficult to get all the developers inside Oracle to change direction. Everything they were reading said that Microsoft and client/server would go on forever. I got the same looks that Galileo got when he told people the earth revolved around the sun. ‘Sure. Whatever you say, boss.’ ”
For the first year, most of Ellison’s Internet evangelizing was confined to technology—the database and tools, the things that he knew and cared most about. But it began to dawn on him that the impact on Oracle’s struggling applications might be even more profound. Ron Wohl took only a little convincing. He says, “In the applications business, you can never catch and pass someone who has dominant market share unless something else changes. If you do the same thing as a competitor who’s ahead of you, you’ll never catch up. That’s the position we were in with SAP.”
Ellison believed that Oracle was so far behind in client/server that the only way to win in applications was to change the game. There were two reasons why Oracle might be able to pull it off. The first was the increasing number of customers that were discovering the difficulty in making client/server systems work effectively over a wide-area network. Client/server might be all right for departmental use, but for any company that wanted to unify its operations over a number of different sites, it was a nightmare. After a decade of client/server as the dominant computing architecture, at least some disillusioned customers were ready to hear about alternatives. The second was that, quite simply, Oracle had much less to lose by betting everything on the Internet. If SAP had done the same and had failed to make the technology transition gracefully, it would have destroyed a wonderful business for no good reason; it’s the dilemma of all incumbents and the reason why the short history of the software business is littered with the bones of once successful firms that failed to adapt to new paradigms in time.
It therefore made sense to Ellison to devote an increasing amount of development time to working out how to make a Web-based, servercentric version of 10.7, even if one of the consequences was to deny 10.7 Smart Client the resources needed to debug it properly before its release. Ron Wohl says, “From 1996 we started moving a small number of key technical people onto the underlying architecture, then a larger number of people, then gradually the whole division. We ran the risk that b
y pouring energy into the Internet version, we were not making the improvements we needed in the client/server version.” For Ray Lane, it was just another example of Ellison and Wohl’s failure to understand the applications business. As far as he was concerned, customers weren’t interested in esoteric architectural issues: what they wanted was reasonably stable software that had all the functionality they needed to run their operations. The fact that 10.7 Smart Client was a couple of years late to market was bad enough; the fact that it still wasn’t much good was intolerable.
What’s striking is that there was so little consensus between the two men running Oracle as to what the applications strategy should be. If Ellison had worked hard to convince the development organization that Internet computing offered a better way than client/server, either he failed to let Lane in on the secret or Lane wasn’t listening. The idea that there was a trade-off between getting 10.7 Smart Client right and Oracle being first to the Internet is not one that Lane has much time for. He claims that when the Internet version of 10.7, known as 10.7 NCA (network computing architecture), came out in early 1998, it wasn’t any better than the client/server release that had been rolled out five months earlier. He says, “NCA wasn’t what it was sold as. We said it was browser-based, but you had to have Java code running on the client or it wouldn’t work.” The only reasonably reliable variant of 10.7, Lane maintains, was the unglamorous character mode version that ran on green-screen dumb terminals and that few customers wanted. While Ellison was busy talking up the brave new world of Internet computing that Oracle was ushering in, one that would solve all the problems with applications, Lane says that customers were still having to live with a very different reality.
In fact, although flawed, 10.7 NCA was a considerable technical achievement, given the compromises that had to be made because of Java’s immaturity. Lane’s suggestion that running client-side Java code meant that it wasn’t really Web-based is technologically naive. Even today, the E-Business Suite runs Java on the client for things such as expense reports. NCA was essentially a three-tiered architecture for delivering Web-based applications made up of an Oracle database engine running on a powerful UNIX server, a lightweight Java application on the client, and a midtier “forms server” to provide the gateway between the two. Among other things, users discovered that since the forms technology now ran on the server, patches to fix bugs could be delivered without having to relink and regenerate customizations. Most important of all, according to Mark Jarvis, who took over as head of marketing a month after the launch, the existence of 10.7 NCA meant that “everywhere Oracle’s sale’s force went, it was pushing the idea of the Internet as being about business, not just a consumery thing—we were educating customers long before the competition.”
Where Lane has a point is that for any customers who had started with 10.7 SC, upgrading to NCA was a nightmare, whereas those with the character mode release found it relatively easy. Unfortunately, communication between development and the field was so poor that customers received little worthwhile advice about which version they should buy. As Ron Wohl admits, “Development must always move ahead of the sales and consulting organizations. It means that half the time they are implementing one product while we are already building the next one.”
If shifting the emphasis away from client/server to the Internet was controversial within Oracle, the issue reached boiling point with Ellison’s decision to halt all further work on client/server in early 1998. Having reluctantly become “Mr. Applications,” Ellison threw himself into the task. One of the problems was that Oracle was now trying to support three different versions of its application suite. Ellison says, “We were the only vendor supporting three different modes of operation: terminal, client/server, and Internet. Life isn’t hard enough—we had to have three fucking versions? The client/server version worked poorly on a wide-area network and was nearly impossible to maintain because the applications code was on hundreds or thousands of desktop PCs. Our so-called Smart Client version was not smart. It was unnecessarily complex, and I wanted to dump it.”
Ellison was desperate to move ahead—“I decided to get out of the client/server applications business just as fast as possible.” He wanted Release 11 to ship as soon as possible to expunge the memories of 10.7, and this time he wanted only one version. If Ellison had his way, 11.0 would be engineered for the Internet only. Ellison was certain that Oracle could not succeed if it had both an offensive and a defensive strategy. The analogy he used was taken from the Six-Day War, when the Israelis had been willing to bet everything on taking out the Egyptian and Syrian air forces on the ground even if it meant leaving their own airspace un-protected. Lane strongly disagreed. It wasn’t that he was against Oracle’s trying to get ahead of the field by embracing the Internet, but he was opposed to leaving in the lurch customers who had signed contracts for client/server software. His fear was that if Oracle was seen as dictating to the market a lot of customers would simply walk.
Ron Wohl comments, “The idea of moving to the Internet was a no-brainer. The hard decision was cutting off client/server and committing ourselves totally. That’s why 10.7 NCA was so important. It would give us references. People had to believe that the Internet version of our applications worked. That allowed us then to go to the next stage and say, ‘Guess what, this Internet technology not only works, but it’s so superior to client/server that we’re not even going to offer client/server.’ We were saying that when Release 11 comes out five months after 10.7 NCA, it’s one thing and one thing only: Internet.”
Ellison is convinced that Lane mounted a serious challenge to his Internet-or-bust strategy. He says, “Ray didn’t want to bet the farm on Internet applications. He thought it was risky to the point of being crazy. He thought that we should do client/server too. But I’d already decided that the Internet applications were our best shot at being successful, and we didn’t have the resources to do both.” In early 1998, soon after Ellison had taken charge of applications, he was briefing some consultants when he received a phone call from Ray Lane. Lane had called a meeting of his top 150 sales managers. He wanted Ellison to come down and answer some of their questions. Ellison remembers, “I got a call from Ray at about midday. He wanted me to come and listen to what the field had to say about giving up on client/server. And they wanted to tell me what customers were saying. When I got there I was treated to what seemed to me to be a carefully choreographed show. The message was clear: our customers wouldn’t tolerate being forced away from client/server and pushed to the Internet. ‘Can you guys just tell Larry that what he’s doing is crazy and he’s putting the entire company in danger.’ I had heard it all before. Relational database will never be commercially viable. The customers want client/server. Blah blah blah. They were mistaking the present for the future. It’s the worst mistake a tech company can make. Client/server was dead, and the people in the room would figure that out at the funeral. By then it would be too late for Oracle to change course. We had to change to Internet applications now—before SAP and the rest figured it out. But Ray was having none of it. I couldn’t convince him or his people.
“This meeting came at a time when I was feeling a bit insecure in my position at Oracle, and if push came to shove, I wasn’t sure whether the board would support Ray or me. Maybe it was paranoia on my part, but the meeting seemed like the beginning of a revolution with me playing the part of Louis XVI. At the very least, Ray was trying to split the company down the middle—his field guys versus my development guys. When that dawned on me, I realized I had to do everything I could to keep a split from happening. So I changed tacks. I carefully and patiently listened to all the field people’s comments and said, ‘You know, you guys are right. I promise we’ll go on doing client/server if we need to.’ But I knew we’d never, ever need to do a client/server version, and I never planned to build one. But I left the meeting letting them believe they had convinced me. They hadn’t. I knew I was right and they were wrong.”
>
Ellison concluded that Lane had attempted to set him up. “I think Ray was trying to do two things: He was trying to make sure that he got a client/server version of the applications. And he was making a power play for more control within Oracle. I was astonished that Ray wouldn’t support my decision to kill client/server and put all our resources on Internet applications. He was fighting a technology decision that was the key to the company’s future. This wasn’t simply about our applications. This was about predicting the future of computing. We needed to move everything—our tools and our database as well as our applications—to the Internet as soon as possible. What was Ray thinking? First he wants to build industry-specific best-of-breed application suites; then he wants to run application development; now he wants to set the overall technical direction of the company. What the hell is going on? That’s my job.”
Lane’s version of the episode is at odds with Ellison’s. He says that the issue of moving away from client/server had not been that important. “It was the same old story that the applications didn’t work and they weren’t competitive and how Ron Wohl was a joke.5 I had about three hours of trying to respond to this and telling them that Larry was in applications now. He is trying to fix these things, guys, give him some time. Finally, I said, ‘You know, guys, you’re talking to the wrong person. You need to talk to the head of engineering, and that’s Larry.’ So I called Larry and said, ‘Come down here.’ It wasn’t a setup at all—how can you set up the CEO? I just wanted him to answer some questions and tell them he was working on it. That’s all. I actually thought he did a great job of handling them.”
Softwar Page 20
