Ellison says, “Ray had done a great job in the U.S., so when he asked for more responsibility I gave it to him. The problem was I gave him too much, too fast. Ray went from running U.S. sales to running global sales in October 1993. That was mistake number one. I removed Geoff Squire from running Oracle Europe. That was mistake number two. Then I gave Ray customer support. That was mistake number three.” The rapidly expanding consulting organization run by Robert Shaw also reported to Lane. When Lane had arrived at Oracle, his position, if anything, had been slightly beneath those of Jeff Henley and Geoff Squire.
Lane, however, bears no responsibility for the firing of Squire, which Ellison now so bitterly regrets. The two men got on well from the outset despite Ellison’s having (typically) failed to explain clearly to either who was the other’s boss. Squire, who had moved back to London after the company had stabilized, says that he would have been happy to work for Lane, knowing that management of global sales would at some time revert to the United States. But Ellison had other ideas. Meeting Ellison for what he expected to be a routine session over breakfast in a London hotel, Squire kicked off his “to do” list with some senior appointments he had been waiting for Ellison’s approval on. Ellison said, “Ah yes, that’s what I wanted to discuss with you, we won’t be needing them. Ray is taking over International.” Squire says: “I then asked him what he wanted me to do. He didn’t say anything, and it was about five to ten and I had another meeting to go to. He looked uncomfortable, so I said, ‘You’re surely not . . . you don’t want me in the company?’ He nodded and I said, ‘That’s ridiculous,’ and I left him there. After that, it got silly with lawyers.”
Looking back on it now, Ellison says, “Removing Geoff Squire from Oracle Europe was the mother of all the management mistakes I made that year. Geoff was probably the most talented field executive we’ve ever had at Oracle. He proved just how good he was by building Veritas out of an acquired amalgam of tech company flotsam and jetsam. Geoff is an extraordinarily gifted guy—and I fired him. Ugh.”
Within two years, Ray Lane was not only very clearly Ellison’s number two but was in practice running far more of the business than his boss. From Ellison’s point of view, it meant that he was able to spend most of his time doing what he liked best: leading the developers in the database and the tools groups. It was also good timing because Ellison’s cofounder Bob Miner had quit his development role at the end of 1992. He was unsuited to managing the big teams of engineers that Oracle by now employed and was fed up with the unrelenting pressure to meet product deadlines. Things had come to a head at one of the Monday management committee meetings over the release date of Oracle 7.1, and Ray Lane had banged the table in frustration.
Ellison recalls, “In the meeting, I asked Bob to commit to a date for Oracle 7.1. He wouldn’t provide a date. I kept asking. Finally Bob got angry and threw a crumpled-up piece of paper in the middle of the table and said, ‘Okay, there’s your fucking schedule.’ The room went silent. After a long pause Bob said, ‘Everyone is working their ass off. It’s done when it’s done.’ That evening Bob called me at home and said, ‘Look, I’m going to make it easy for you. I’m going to let Derry run database development.’ Bob knew I would never take him out of his job—so Bob did it himself.”
A year later, Miner was diagnosed with lung cancer, and a year after that, he was dead. Ellison says of him, “Oracle changed a lot of people, but not Bob Miner. He never bought a new house. He never bought a fancy car or boat. He never ever changed. He never got swept away by ambition.”
The departure and death of Miner affected Ellison personally, but without him, the little group Miner had taken off to San Francisco disbanded and returned to Oracle, mainly to the database group. One of them was Ed Screven, who had previously worked with Jeff Walker. “Bob was an incredibly smart guy with great technical intuition. We were trying to build a kind of next-generation application execution environment. We had a lot of good ideas and it was fun, but we didn’t make much progress. It was just way too ambitious.”
As for Ellison, despite a fair amount of turmoil in his personal life (see Chapter 20) and having survived a couple of near-death experiences of his own (between the end of 1991 and the middle of 1992, he had managed to break his neck while body surfing in Hawaii and smash one of his elbows when his front bicycle wheel stuck in a disused railway track while he was careering downhill near Napa), he was fully focused on propelling Oracle clear of the pack of other database vendors. His strategy rested on three legs. The first, and most important, was achieving real product superiority over Sybase and Informix, Oracle’s main rivals in the early 1990s. With the release of Oracle 7 and, in particular, Version 7.1 in 1993, Oracle had, for the first time in several years, unambiguously the best database on the market. The second was in some ways a repeat of what had worked for Oracle in the 1980s when Ingres had been the target. Ellison was determined to exploit the strength of Oracle 7 by building market share no matter what the cost.
Ray Lane believes that Ellison’s obsession with growth may have been one of the factors behind his sacking Geoff Squire. Although Europe was extremely profitable, it was not growing fast enough for Ellison. The country managers appointed by Squire to run subsidiaries, such as Michel Rocher in France, had considerable autonomy, and some of them preferred slow growth and big profits. Lane says, “Larry was constantly saying he wanted better numbers out of France, out of Europe. He was saying that we were losing market share to Sybase every day in France and he didn’t care how profitable France was. I told him, ‘Larry, if we do this, ramp up growth, we’ll need to hire more salesmen, really saturate the market, but we’re going to have a profitability problem.’ He said, ‘I don’t give a shit about profits. I want to beat Sybase.’ So that’s what we did.”8
It didn’t take long to work. Lane remembers, during a trip to London in early 1994, Ellison getting him out of bed at three in the morning to deliver the news that Sybase had badly missed its quarter. “He just said, ‘It’s over.’ He was the happiest I’d ever known him. They never, never recovered. Then it was Informix. I remember we broke Informix in about a year and half. We had more salespeople, we went for their key accounts, and we made sure that we never lost on price.” Tellingly, the bonus scheme that Ellison, Lane, and Henley shared in was based on one thing: market share. If Oracle was growing at twice the rate of Sybase, the bonus would be doubled. Lane recalls, “The year Informix hit the wall the bonus got multiplied times thirty-three, but of course, the board couldn’t let that happen. Larry has to have an enemy. Larry’s the most focused individual I’ve ever seen when he has something to beat.”
The third leg of the strategy was for Oracle to be increasingly seen as more than just a maker of databases. Ellison’s interest in massively parallel processing (MPP), or clustering technology as it was also known, was as strong as ever. It had helped deliver the crushing benchmark performance figures that Oracle’s marketing machine had blitzed the opposition with, but Ellison believed that it might have another application that could take Oracle into a very different world from the corporate computing engine rooms that had been its sole habitat up to that point.
In the early 1990s, there was growing excitement about the perceived convergence among the telecommunications, media, and computing industries. The idea was that the telephone and cable companies would build incredibly high capacity fiber-optic networks, replacing old twisted copper wire and primitive one-way broadcast cable systems; the computer firms would come up with ways to move data from one end of the network to another at high speed and convert bits of information into high-quality streamed sound and pictures; the television companies and the film studios would digitize their huge programming libraries and develop interactive and transactive content with retail and financial services partners; the killer applications, many believed, would be video-on-demand (VOD) and home shopping. The whole caboodle went by the name of the “information superhighway,” and everyone, it seemed, wa
nted to be a part of it. Including Larry Ellison.
Ellison reckoned that he had at his disposal a number of the pieces that were needed to make the information superhighway, and VOD in particular, possible. The two assets Ellison controlled were the leading relational database management system and a majority stake in a Silicon Valley–based computer hardware company called nCUBE, which he had acquired in 1988 for around $11 million of his own money (Bob Miner had put in about $5 million). Ellison maintained that Oracle should never go into hardware itself because it must always be seen to be platform-agnostic. nCUBE had been started five years earlier by some engineers from Intel, and it was an early leader in building massively parallel computers—relatively compact machines with multiple processors and the ability to calculate far faster than conventional mainframes.
Ellison hoped that one day an ultrareliable version of the Oracle database would run on the nCUBE boxes. It was partly his belief in MPP as the way of the future that had resulted in the rewriting of Oracle 6 with all the consequent problems of lateness and chronic unreliability. Unfortunately the nCUBE machines were very tricky and time-consuming to program, which made them unsuitable for the rough-and-tumble of corporate computing. But Ellison realized that it might be possible to use them for the much more specialized task of serving up streamed video.
What attracted Ellison to video-on-demand was not just the possibility of being able to sell more software and finally make some money from nCUBE but the opportunity to link Oracle with something that everybody was talking about and that was more glamorous and consumer-oriented than Oracle’s usual business. Ellison says: “VOD had become very fashionable.9 The technical problem presented by VOD was moving very large amounts of data, all those bits that make up the movie images, off disks and onto the Net fast enough. The performance bottleneck was the video server computer’s memory system. But if you have multiple memories, as you have with a massively parallel machine, then you have all the bandwidth you need for VOD. Problem solved. So we set off to build the software for a massively parallel video server.”
In 1993, Ellison co-opted a team of some of his brightest young engineers to solve the technical problems of building a media server that could handle video, audio, and text on the nCUBE platform. Within a few months, they succeeded. How much the software had to do with Oracle’s database technology is debatable, but at least the shared-disk approach to architecture that had made Oracle 6 so hard to finish finally came into its own. By January 1994, the technology was sufficiently stable to be rolled out at an event hosted by Walter Cronkite at the CBS studios in Hollywood.
Ellison was already experiencing a degree of limelight that didn’t normally come with selling databases. He met entertainment moguls such as Michael Eisner of Disney, Rupert Murdoch, and Barry Diller, and also made his first appearance on the cover of Fortune magazine. The accompanying article by Alan Deutschman, gratifyingly entitled “Software’s Other Billionaire,” was in marked contrast to the kind of press that Ellison had been getting only a couple of years earlier. Best of all were the pictures showing Ellison in the Japanese garden of his home in Atherton surrounded by lots of little television sets. The message couldn’t have been clearer or better for Oracle: here was the man who was making the dream of video-on-demand into reality.
Was it all a dazzling digression from Oracle’s real business, a bit like the network computer that was still to come? Marc Benniof strongly disagrees: “With both the network computer and the video server, you can see a very important part of the way Larry thinks about things. He will create a strategy that elevates Oracle and also destabilizes a competitor in one movement. It’s like the martial arts of business. He’s very good at this. The video server strengthened Oracle because it made people much more aware of Oracle, the Oracle brand, and Larry the visionary. After the Fortune article there were a bunch of other pieces all saying the same kind of thing. It not only drove people to Oracle, all the competitors were going crazy saying, ‘Gee, what should we do?’ Microsoft had to respond and so did Silicon Graphics, which was very hot in those days, and the other database companies. They don’t even know what’s happening to their own psychology when Larry is out there doing his stuff.”
The glitzy presentation of Oracle’s all-singing, all-dancing VOD technology at CBS didn’t go quite as planned, but not because of the gremlins that traditionally conspire to undermine carefully rehearsed tech demos. At 4:30 on the morning of the rollout a powerful earthquake measuring 6.7 on the Richter scale struck the Northridge area of greater Los Angeles, destroying buildings and collapsing freeways. Ellison was in his hotel. “People were pounding on my door telling me to get out of the hotel because of the earthquake. I told them to go the hell away so I could go back to sleep. I had already been awakened by a huge crash in my room followed by what sounded like a series of gunshots coming from the bathroom. I didn’t care. I wanted to be well rested for the presentation the next morning. I went back to sleep. The next morning I discovered what caused the noises in my room. The ‘gunshots’ were ceramic tiles exploding off the bathroom wall as the building swayed back and forth. The big crash was the armoire with the big-screen TV falling face first onto the floor.”
Maybe it was also an omen. The demo had to be postponed because Cronkite was pulled out to work on the earthquake. Unlike the television in Ellison’s room, VOD went out with a whimper rather than a bang. It wasn’t, however, because the technology didn’t work. Time Warner ran a big pilot for a couple of years down in Orlando, Florida, while Oracle partnered big telephone companies in similar pilot services on both sides of the Atlantic. One was with British Telecom, involving two thousand paying customers in Ipswich, which ran for a year, until the end of 1995. Oracle provided the software, nCUBE the servers, Apple had been persuaded by Ellison to build a set-top box based on the Macintosh computer, BT ran the high-speed network, and a group of media companies from the BBC to some of the Hollywood studios contributed content.
The findings of the experiment were pretty uniform. The technology scaled perfectly, as Oracle had promised it would, although the nCUBE boxes were expensive because of their proprietary processors. People also liked the service a lot and were prepared to buy some of their television on a program-by-program basis (in Ipswich, the most popular feature was the chance to watch favorite soaps a day before they were broadcast, thus enabling subscribers to be one up on their friends). Unfortunately, they weren’t spending enough, either on entertainment or on the primitive online shopping that was also provided, to give the telcos any confidence that they would get a return if they spent billions building out the new networks and running fiber to the doors of their customers.
Ellison says, “It’s really embarrassing looking back on it. We all believed this was going to happen. Bill Gates even wrote a book about it. I thought I was so smart figuring out how to use a massively parallel computer to solve the VOD technical problem. Unfortunately, the barrier turned out to be economic, not technical. Consumers were unwilling to pay very much for VOD movies, and the market for training videos and shopping services was too small to make up for it. It didn’t matter how good our video server was, it just didn’t make economic sense to spend billions building a broadband network to everyone’s home.”
Three other developments also explained the telcos’ failure to push ahead. One was an emerging technology called digital subscriber line (DSL), which offered the promise of turning standard paired copper wire into a much fatter pipe. While the bandwidth would still be much less than that of fiber, advances being made in digital compression offered the possibility of soon being able to deliver broadcast-quality video through the existing network. BT said it might mean the difference between spending $3 billion and well over $20 billion. The second was the imminent arrival of digital television with the capacity to deliver more than five hundred video streams, a good many of which could be used to offer movies with staggered start times on a pay-per-view basis. Near-video-on-demand was no
thing like the real thing, but the telco bosses were worried that a lot of people would think it good enough.
Finally, there was the arrival of the Internet as a mass-market platform, thanks to the creation of the World Wide Web by Tim Berners-Lee and Marc Andreessen’s point-and-click browser. While the Web still needed big bandwidth to provide video of acceptable quality, it was a fundamentally different approach to the proprietary networks that the telcos had had in mind. And while the Internet might not yet be a good way of distributing movies, it was likely to be much better at delivering most of the other online services that the telcos had hoped to earn commissions from. By the end of 1996, almost nobody was talking about the information superhighway anymore. Video-on-demand might have been only resting, but nobody could say when it would be revived.
But for Larry Ellison, the whole interactive TV thing had been a wonderful ride that had cost very little money and had fundamentally changed the public perception of Oracle. At the beginning of the 1990s, hardly anyone had heard of Oracle except for IT managers. By the middle of the decade, if it wasn’t exactly a household name, it was widely recognized by technology investors and readers of the business press. Ellison had set out to stop people talking about Oracle and other database companies and get them talking about Oracle and Microsoft. “Oracle was now perceived as a software heavyweight, ranked just behind Microsoft and IBM. Customers who bought our products could trust that we were going to be around for a long time. We were a survivor. The same could not be said for Sybase and Informix. And if customers don’t believe you’re going to survive—you won’t.”
Softwar Page 15