Softwar

by Matthew Symonds

  But as usual, Ellison didn’t stick around for long. By 1976, he had moved to a firm called Precision Instruments (soon to change its name to Omex), that, like Ampex, was also working on a mass storage system. The difference was that for the first time Ellison had a title—vice president of R and D—and some seniority. “I had always believed that at the top of these companies there must be some exceptionally capable people who make the entire technology industry work. Now here I was, working near the top of a tech company, and those capable people were nowhere to be found. The senior managers I saw were conformist, bureaucratic, and very reluctant to make decisions. Much to my surprise, I gradually became convinced that I was better at solving problems and making decisions than they were. Or at the very least I was willing to make a decision to do something, while they seemed paralyzed by endless analysis and fear of making the wrong decision.”

  Ellison suddenly began to think that maybe he could run a company. But he wasn’t motivated by either great ambition or the vision to do anything in particular. Having his own company would simply allow him to go on with his life but with more control over it and with better rewards. “I thought that if I ran a small consulting company, I could work on a lot of interesting projects and make more money. The key was finding enough projects that were ‘hard’ enough. The ideal ‘hard’ job is one that is generally perceived as being difficult but can quickly succumb to a bit of cleverness. All we needed were a few really smart people who could do these so-called hard jobs quickly, and we would make a lot of money while not working all that hard. That had been my favorite approach to work from the beginning.”

  The job at Omex gave him his opportunity. Omex was preparing to put the contract for writing the software for its mass storage system out for bidding. “I said I knew some super programmers who could save the company a lot of time and money. Do you want to me to ask them to put together a bid on the contract? My boss, the CEO, said, ‘Sure.’ ” The team that Ellison had in mind included two of his former colleagues from Ampex, Oates and Miner. The three of them formed a company—Software Development Laboratories—and they won the contract from Omex with a bid of $400,000, a third the price of the next lowest bidder. Ellison would stay at Omex to design and oversee the project while Miner and Oates wrote the software with the help of a hotshot young programmer named Bruce Scott.

  When it came to negotiating the ownership of SDL, Ellison says, “Initially, I proposed splitting the company evenly between Bob, Ed, and me. Then I went home and thought about it. The company had been my idea, and so far I had done all the work. So I changed my proposal. I wanted sixty percent, while Bob and Ed would get twenty percent each. We agreed that this division of ownership was just a starting point and that both Bob and Ed could earn additional stock so long as they performed well. Bob did a great job, and his ownership percentage was increased to thirty percent. Ed went through a very messy divorce, which affected his performance, and he ended up leaving the company. Eventually Ed returned, but his ownership was reduced.” Over the years, as Oracle grew, went public, and became a blue-chip stock held by many institutions, Ellison’s share has inevitably been diluted, but he has always been acutely aware of the value of his stock both in financial terms and because of the power it gave him within the company.

  Alongside doing the work they were doing for Omex, the SDL partners, with Ellison at last installed as president, began thinking about what sort of company they wanted it to be. For Ellison it seemed obvious. He says, “I wanted to get out of the consulting business. Consulting proved to be much more work than I ever imagined. My ‘hard problems, cleverly solved’ business model did not scale up beyond a few people, proving I was not nearly as clever as I thought. I was working eighty hours a week—at least. We were making a lot of money, but we were working insanely long hours. The closest I got to Yosemite was the Ansel Adams poster on my office wall. So we decided to abandon the consulting business and go into the software product business. A software product offered the ultimate leverage: build it once and sell it over and over again.” The problem was deciding what it was they would make. Ellison says, “When I was working for Bob Miner at Ampex, I built a database management system for the PDP-11 minicomputer. It was a CODASYL database, basically a ripoff, excuse me, ‘modeled after’ Cullinet’s successful IDMS mainframe database. Cullinet had been quite successful competing against IBM in the mainframe database business. Surely there was a market for minicomputer databases as well.”

  The SDL guys scoured the technical literature about databases for new ideas. Ellison says, “The first time I heard about a relational database was some years back at a meeting with the CIA. Since then I had read a series of technical papers by [Edgar “Ted”] Codd, an IBM fellow, about relational database theory. But Codd’s papers were much too theoretical to be used as the basis for a product. Then IBM Research starts publishing this series of papers by Don Chamberlain about a prototype relational database they were building called System R, including a complete specification for System R’s Structured Query Language, or SQL. After a lot of careful reading and rereading of the System R papers, I decided we should throw out the work I had done on the CODASYL database and use the IBM papers as an architectural blueprint for our new database product. The opportunity was huge. We had a chance to build the world’s first commercial relational database. Why? Because nobody else was even trying.

  “The other relational database projects were pure research efforts. They were spending little to no time on overall system performance and reliability, both of which were essential for a successful commercial database product. If we could build a fast and reliable relational database, we would have it made. At the same time I thought it would be smart to avoid the very competitive and highly conservative IBM mainframe market. So we decided to build our database for the minicomputer market. Bob thought that the relational database strategy was risky, but he left the final decision up to me. I thought that relational was clearly the way to go. It was very cool technology. And I liked the fact it was risky. The bigger the apparent risk, the fewer people will try to go there. We would surely lose if we had to face serious competition. But if we were all alone in pursuit of our goal of building the first commercial relational database system, we had a chance to win. But we had to be first to market with the new technology. This was the first in a pattern of apparently high-risk decisions I made throughout my life at Oracle. But I only ever picked the high-risk approach when I thought that it would increase our chance of winning and our share of reward.”2

  In the relational model, information in the database is viewed as a number of linked two-dimensional tables consisting of rows and columns. Each table has a specified set of named columns, while each column name is distinct within a particular table, although not necessarily between tables. Within a particular column, each entry must be of the same type. Unlike the hierarchical approach, no predetermined relationship exists between distinct tables. By combining the data items in related columns, it is possible to find information that would either be hidden in a hierarchical database or would take far longer to find. Relational users don’t need to understand how stored data is represented in order to retrieve it. What was so exciting about SQL was that it was the only language needed for working with relational databases. It could be used not only for queries but also for data manipulation, data definition, and the setting of access or privileges for groups of users. No wonder Ellison thought it was “cool technology.”

  Given that IBM was doing most of the important work in the relational area, apart from the CIA-funded Ingres project at Berkeley under Mike Stonebraker, what made Ellison think that he and his tiny team could come up with a commercially viable product before anyone else? “First, conventional wisdom had it that relational databases couldn’t be made to run fast enough to be commercially viable, so people didn’t enter the race. Second, IBM feared that introducing a relational database would destabilize its large existing IMS [Information Manag
ement Systems] database business. Third, even if IBM did deliver a relational product, it would probably be for mainframes. Since we were targeting the minicomputer market, we didn’t need to worry about IBM—at least not in the beginning, when we were most vulnerable.”

  The money that was coming in from Omex basically paid for SDL to create a first version of what would become the Oracle database. This prototype didn’t really work, but SDL was creating intellectual property for itself, changing from being a consulting firm into a software company at a time when there were still very few pure software companies around, as opposed to hardware companies that wrote the code to run their boxes. When Ellison looked around, Boston-based Cullinet, with its market value of around $100 million, seemed like the company to emulate. “Cullinet was our role model. We never thought we could get as big as they were because they had an expensive mainframe product and we were focused on these cheap little minicomputers.”

  After completing the work for Omex, doing just enough to get paid for its software despite the fact that the firm’s thin-film laser storage hardware still didn’t really work, SDL changed its name to Relational Software, Inc., or RSI, and moved into some smart new offices at 3000 Sand Hill Road near Palo Alto, the address of the top venture capital firms in the Valley. RSI’s first customer turned out to be the CIA. The CIA had been interested in the relational model ever since Codd had first expounded his theories and had avidly followed the work IBM’s System R research group had done to make SQL into a key to unlock information from a database that could be used by any half-competent user. Unfortunately, IBM wasn’t yet interested in making anything that might undermine IMS, so the CIA had to find somebody who might be interested in coming up with a commercial project. Dave Roberts, the man whose job it was to find a potential vendor, soon tracked down RSI. Two things that encouraged him to bet the agency’s money on the tiny firm were the discovery that Bob Miner, who had worked with him on the Ampex terabit memory project, was involved and that RSI was committed to SQL and therefore had IBM’s stamp of credibility. The other thing that appealed to the CIA was that RSI was building its database to run on the PDP-11 minicomputer. Ellison says: “Digital’s PDP-11 minicomputers were heavily used in government, especially within the intelligence community, because the computers are small enough to fit into airplanes and submarines.”

  The money that came from the CIA enabled RSI to finish work on the first commercial release of the Oracle database. Ellison recalls that the moment it actually worked brought the kind of thrill the Wright brothers must have had when their plane took to the air. “The first version of our database was called Oracle Version 2. I didn’t think anyone, not even the government, would buy Version 1 of a database from five guys in California. Unfortunately, the Version 2 label didn’t change the fact that we were experiencing those very serious performance problems predicted by conventional wisdom. We just couldn’t make our database go fast enough. We tried one thing after another. Finally we got a breakthrough that delivered the tenfold performance improvement we needed for commercial viability. In our final test we ran faster than the CODASYL system, which was then considered the fastest PDP-11 database.

  “I was stunned by the final test results. I had never thought about the consequences of failing, but I had never thought about the consequences of succeeding either. On my way out that evening I ran into Tony Spoor, our auditor, in the elevator lobby. I was a bit light-headed and euphoric, still trying to fully grasp the import of our performance test results. I spontaneously decided to explain what it all meant. After laboriously rattling off the details of our performance test results, I said ‘Tony, do you know what this means? Do you? Do you know what this means?’ He said, ‘Ugh . . . no.’ I said that it means Oracle’s worth at least $100 million. Tony’s lips parted slowly and his jaw lowered. He did not respond, but his look said it all. He thought that after all the late nights and weekends my mind had finally snapped like a dry twig. This hundred-million-dollar valuation I just proclaimed seemed so far from reality that he thought I should be sedated and put into a padded cell so I wouldn’t hurt myself.”

  Ellison adds, “It gradually dawned on me that this thing was going to take on a life of its own. Not because we were going to grow to be a big company. I think we all had the fantasy that however successful our database product, there would never be more than fifty of us, and customers would be sending us checks and we’d be sending them software. Even with just fifty people on board, I thought that my management ability would be stretched to the limit. I remember when we hired our first salesman, Bob Preger, who came from a mainframe database company called Software AG, Bob Miner came in and asked me, ‘What do we need a sales guy for?’ I smugly replied, ‘To sell more of our software, perhaps.’ That’s an inside glimpse of our top management team at work discussing the expansion of our distribution capacity. I knew we needed to build a sales organization, although I certainly had no idea how one worked. Bob, however, remained unconvinced as to the need for salespeople. He just said, ‘Okay, we’ll try one.’ ”

  The only way to convince people that RSI had a product that worked—albeit, according to Ellison, “not all that well”—was to demo the new database system to anyone who could imagine a use for it. Ellison had to hit the road and sell. “The news about our hot little database traveled around the intelligence community pretty quickly. I’d make a presentation at one agency on Monday, which would result in a phone call on Tuesday to visit another agency on Wednesday. One week I flew to and from Washington, D.C. three times. In a little over six months’ time we had won several deals—the CIA, Navy Intelligence, Air Force Intelligence, and the NSA [the highly secretive National Security Agency].”

  But as so often happens with young companies, just at the moment that things appeared to be taking off, RSI came perilously close to running out of cash. “Once we started making database sales, we decided to stop doing consulting work. We needed everyone working on finishing the bloody product.3 So the consulting money stopped coming in. All we had was $235,000 in the bank, and that didn’t go very far even in those days. But the real killer was that I didn’t have the slightest idea about how much time could elapse from the moment the government says they’re going to buy and the moment you get the cash. I thought we’d have our money in sixty days, max. Not even close. When the government wants to buy anything, they have to adhere to a congressionally mandated process, which includes the publication of the pending purchase in the Commerce Business Daily so other firms can offer a competitive bid or protest the purchase. So, after you’re told you’ve ‘won’ the deal it can still take six to nine months before you actually get paid. No consulting money was coming in, so we were going to run out of cash. The solution was simple: we had to stop paying some people, starting with me.”

  • • •

  Far from finding running his own company liberating, Ellison was working harder than he’d ever imagined, to the extent that it probably contributed to the breakdown of his second marriage, to Nancy Wheeler, a grad student at Stanford, after only eighteen months. “Self-absorption will crush a young marriage, and I was completely absorbed by my own ambition. I was totally immersed in making the business succeed. I neglected everything else. I was writing computer programs, writing documentation, answering customer support calls, giving training classes, and sending out marketing literature. The very last thing I would do every evening was stamp a stack of manila envelopes stuffed with this spiral-bound Oracle introductory manual I’d written. Everyone and anyone I spoke with on the phone that day would get one. I’d arrive home around midnight most evenings. Nancy eventually tired of this routine and left. She was the second wife I managed to drive away. The first left because I refused to work enough. The second left because I worked too much. Both cases were caused by my self-indulgent, without-compromise mode of living. My personal life had fallen apart, and work was the only thing left that mattered. So I worked even more—until the hours of the day ran ou
t.”

  The long hours being put in at RSI were noticed by another occupant of the building, Don Lucas, a doyen of the Valley venture capital community. Lucas recalls, “I was one of the developers of 3000 Sand Hill Road, and I had a nice corner office on the second floor. I was keeping some pretty odd hours myself, but I noticed that whenever I left for home the lights in this particular spot in the complex were always on. So I finally went by and asked them what they were doing and I met Larry and Bob Miner and Barbara [later to be the third Mrs. Ellison], the receptionist. And we just got talking. They clearly didn’t have a lot of money, and they weren’t very sophisticated in the ways of corporate America. The first thing they asked me was if I’d go on the board. So I said, ‘Sure,’ and I became chairman of the board.”

  Although it was part of Lucas’s job to make equity investments in promising young companies, Ellison and Miner were insistent that they didn’t need money. When they urgently needed to get inside Digital’s new VAX, a sort of PDP-11 on steroids, to ensure that their software could run on the minicomputer that over the next ten years would sweep all before it, rather than push for a stake, Lucas arranged and guaranteed a loan. “They were very energetic, very positive, driven. These guys worked not round the clock but pretty close to it. There was a very interesting relationship between Larry and Bob. Larry was the visionary, the top salesman, the top cheerleader, the driver, and Bob was clearly a very good software architect.”