2. LE writes: When we implemented our E-Business Suite at JDS Uniphase, it was up and running globally in less than a year. We put the system in for a guaranteed fixed price. JDS’s IT budget is now 50 percent lower than what it was before.
3. LE writes: The $20,000 special edition of the E-Business Suite came to the China market in 2002.
4. LE writes: It has been my experience that people reflexively resist change. Change requires people to rethink the way they work and the way they are organized. When people write down their current processes and the reasons why those processes cannot be simplified, it forces them to carefully and methodically rethink their business. This usually results in business processes being changed, as opposed to our software. Simplified, modernized business processes are at least as important as good business software in delivering efficiencies to the enterprise. When we do turn up a serious shortcoming in our software, we add the required feature in a future version of the product at no charge.
5. LE writes: Oracle’s e-business application software helped BellSouth become number one nationally in high-speed ADSL networking. Unfortunately, our software was heavily modified and extended to meet BellSouth’s requirements. Making major changes to our standard e-business software—or anyone else’s for that matter—is an expensive and time-consuming process.
6. LE writes: One of the oldest and worst habits of the Oracle sales organization was the use of aggressive discounting to get existing customers to buy software before they actually needed it. A customer might be planning a $2 million software purchase in 2004, but the sales force would offer the same software for $1 million if the customer bought in May 2003—the last month of our fourth quarter. Customers quickly figured out that the best time to negotiate for discounts was the last day of our quarters. It has taken years and a lot of management changes before we were able to break this habit.
7. LE writes: The largest component of that $800 million was internal Alcoa labor cost. Oracle consulting and software were less than 10 percent of the cost of the project.
3
THE WAR ON COMPLEXITY
Larry Ellison says he’s happy only when everyone else thinks he’s wrong, when he’s “walking way out to the end of the limb and then jumping up and down.” I don’t think that’s bravado. The core of his business philosophy is that you can’t get rich by doing the same thing as everyone else. “In 1977, everyone said I was nuts when I said we were going to build the first commercial relational database. In 1995, everybody said I was nuts when I said that the PC was a ridiculous device—continuously increasing in complexity when it needs to become easier to use and less expensive. By then it was clear to me that PC-centric client/server computing was a mistake, a misguided model that just distributed complexity. And now they say I’m nuts because I said we could build a complete and integrated suite of application software. I always feel good when everyone says I’m nuts because it’s a sign that we’re trying to do something innovative—something truly new and different.”
Ellison pauses and emits his strange, wheezy, high-pitched sigh of a giggle. “On the other hand, when people say you’re nuts, you just might be nuts. You’ve got to constantly guard against that possibility. You don’t want people saying you’re nuts too often—once every three or four years is good. Any more than that, and you should be worried, because no one’s smart enough to have a good idea more than once every three or four years.”
The E-Business Suite was released at the beginning of June 2000. The suite was controversial right from the outset, and Ellison wouldn’t have been true to himself if he hadn’t gone out of his way to make it even more so by using it to challenge most of the belief system that the software business was founded on. In fact, the E-Business Suite was intended as an assault on the entire ecosystem of enterprise computing. It was an attack on every rival application company because of Ellison’s claim that it was so complete customers could get everything they needed from one vendor; it was an attack on the huge and powerful systems integration industry because Ellison was arguing that gluing together programs from different software firms was an expensive waste of time; it was also an attack on the analysts in the consulting firms whose raison d’être was to identify the “best-of-breed” supplier in each category of application.
Ellison’s advocacy of the E-Business Suite was not based on the belief that it had more features than rival products—in fact, he was ready to admit that in a straight features contest between Oracle’s suite and most best-of-breed assemblies, the E-Business Suite would come off second best. He said, “It may not have one hundred percent of what you want, but it has a hundred percent of what you need. The advantages of out-of-the-box integration more than make up for a few missing nice but not necessary to have features.”
What did he mean? The overwhelming reason, Ellison believed, for the failure of information technology to deliver on its overblown promises was the fragmentation of information that resulted from the computing industry’s addiction to complexity. Expensively assembled best-of-breed enterprise computing systems, in which the cost of implementation might be ten times as much as the original software licenses, were Ellison’s latest target but the origins of his “war on complexity” went back more than a decade.
Ellison started his career as a contract mainframe programmer, and although Oracle had been happy to join the stampede to client/server in the late 1980s, Ellison quickly became disillusioned with it. He says, “We installed lots of client/server systems. But it was a very ugly model. Our applications were installed on thousands of PCs, so if you needed to fix a bug you’d have to change the software on all those thousands of PCs. It was an incredibly labor-intensive, expensive, error-prone nightmare of a process. Worse yet, client/server worked fine on high-speed Ethernet LANs [local-area networks], but it performed terribly on [slower] wide-area networks [like the Internet]. We’d taken a huge step backwards because of this wide-area network problem. Mainframe systems had dumb terminals with a dumb user interface, but dumb terminals worked beautifully on wide-area networks, so you could build an airline reservation system on a single computer with a single database that could be accessed by terminals all over the world. But client/server systems didn’t work on wide-area networks, so you had to use a separate local-area network in every location you did business. Attached to those local-area networks were database and file servers—lots of little servers everywhere, lots of little databases everywhere. Your information got hopelessly fragmented in the process.”
Ellison isn’t short of stories that exemplify the “insanity” of client/server: “I’ll never forget when this reporter from The Wall Street Journal phones me to get my comments on a story he was writing on Grand Met’s [the international hotels and catering group] information strategy a few years ago. They decided that they were going to put a ‘low-cost’ database server computer into every Burger King. ‘What do you think of that, Larry?’ I was stunned. All I could say was ‘They’re putting databases in hamburger stores? What? Are they crazy? They shouldn’t put a database in every store.’ But they were kind of stuck. The only way for them to deal with client/servers’ dependence on fast networks was to put a server computer into every location where they did business, that is, in every location where they had client PCs. And if you think it’s hard to manage lots of desktop PCs, just wait until you try to manage a database server in every location where you do business. Good luck. It’s nearly impossible to manage all that distributed complexity; plus, you so fragment your data you can’t keep track of what’s going on in your business. But all these client/server problems weren’t obvious until you got to the point of trying to run a large network of lots of little database servers in different locations. In contrast, the advantages offered by PC client/server computing were immediately obvious to everyone and quite seductive: better user interfaces; cheaper hardware; easier programming—all that made client/server seem irresistible. But the inherent problems of the client/server model
were enormous. It was a ticking time bomb.”
In fact, the bomb was ticking inside Ellison himself. The day it exploded was September 4, 1995, at the annual conference in Paris called the European Information Technology Forum. It was exactly eleven days since the launch of Windows 95 accompanied by the biggest marketing blitz in history. Bill Gates’s face was everywhere. He was feted not only for being the inspiration behind the most exciting company in the world but for the powerful vision of the near future that he had articulated (with the help of Microsoft’s chief seer, Nathan Mhyrvold) in The Road Ahead—a book all about the coming information superhighway that somehow managed to avoid almost any mention of the Internet. Gates was to be the keynote speaker at the conference, with Ellison scheduled to be the warm-up. Ellison’s speeches are rarely scripted, predictable affairs. On this occasion the slide projector didn’t work, so he just started talking. Did he know what he was going to say beforehand? “Sort of, but not exactly,” he says.
What he thought was important to talk about that day was the “paradigm shift” to the Internet that was coming—similar to the shift from mainframes to client/server—and to express some dissent about the apotheosis of the Microsoft Windows PC, which had clearly taken its toll on Ellison. He says now, “It was really getting on my nerves. It was just such a surreal experience. There was peace in the Middle East and war in Bosnia the same week that Windows 95 was announced, and all the major networks seemed to cover was people in parking lots waiting up all night to get their first copy of Windows 95. It was very strange. Everyone was saying that Microsoft’s PC-centric client/server computing would take over the world and then go on forever. Everything else would disappear. UNIX was going away. Mainframes were going away. There’d be little Windows computers everywhere; little database servers everywhere. Well, we were in the middle of installing all these client/server systems and trying to make them work. And they didn’t work very well. Your information was chopped into tiny pieces, stored in lots of tiny databases, running on lots of tiny PC server computers. This data fragmentation was accompanied by distributed complexity. The PC was already a ridiculously complex device. And Microsoft had plans to make it still more complex. They were developing a new, more powerful operating system designed to run all the increasingly complex applications that were destined for your desktop PC. It was madness. You can’t keep putting applications on the desktop PCs. You have to get as many applications off the desktop and onto servers as possible. And while you’re at it, most data should be moved off PCs and onto servers as well. The PC needs to be much simpler and more appliance-like.”
The big idea in the Paris speech was that even at the moment that the PC’s dominance seemed beyond question, the rise of the Internet was already demonstrating a superior model of computing. Ellison had said, “When you have a networkcentric computing model, you don’t need a device anywhere near as complicated as a personal computer. You can build a multimedia Internet terminal for $400 or $500 [compared to the $2,500 you needed to spend at the time to get a capable PC]. You just plug it into the wall to get power—electrons—and into the network to get to applications and information—bits. And you’re done.” The one mistake that Ellison made was in calling the new “thin” client (as opposed to the “fat” PC client) he had dreamed up a “network computer.” Looking back, he says, “If I’d been really smart I’d have called it an Internet computer. I called it a network computer partly because I liked Sun’s slogan—‘The network is the computer’—and partly because the word ‘Internet’ was not all that well known in 1995.”
More than two and a half years later, when Oracle shipped an Internet version of its previously client/server-based Release 10.7 applications suite, Ellison still hadn’t learned that particular lesson. The reengineered software was christened 10.7 NCA, the initials standing for network computing architecture, and no one was listening. Ellison recalls, “We had to change the name. Network computing architecture got absolutely no traction in the marketplace. So we announced a successor called Internet computing architecture, and sales took off. The big difference between the two? The name.”1
The consequences of the Paris speech were profound. Although Gates, who had followed Ellison onstage, didn’t stoop to reply to the attack on his beloved PC, the media had the bit between their teeth. Ellison was suddenly front-page news. He was actually not predicting the death of the PC (the PC, like the mainframe, wouldn’t disappear, but it would no longer be at the heart of the computing universe); however, that was how it looked as the story gathered momentum. In an interview after the speech, Ellison, enjoying the attention, characteristically added a little fuel to the flames, boldly forecasting that within ten years, network computers would have overtaken PCs in sales. The ensuing publicity for Oracle and Ellison was beyond anything he had dared hope for. Suddenly he was much more famous than he would ever have been as the boss of a company that sold arcane technology to businesses. Now he was locked in a theological struggle for the future of computing with one of the richest and most recognizable people in the world. Oracle, instead of being bracketed with other database firms, such as Informix and Sybase, was suddenly being talked about in the same breath as mighty Microsoft.
It quickly became apparent that the world wanted to know what the newfangled network computer would look like and that Oracle, a software rather than a hardware company, would have to find a way of building one. Mark Jarvis, who now runs marketing at Oracle, recalls, “Microsoft was saying no, this will never happen, it’s a stupid idea, so we were actually confronted with this problem of having something to show people. Larry was a bit responsible for this, everyone thinking of the network computer as a physical device. Later on, as history now shows, the PC sort of became a network computer and so did the cell phone. But back then we had to make something that people could look at.”
The official unveiling of the network computer was at Oracle Open World in Japan, one of the biggest computer shows in the world. Jarvis says, “We had them under cloth at OpenWorld. Larry unveiled them in front of all these people and they looked great, flash, flash, flash, there was a media frenzy. But you know what, they were made out of wood. I remember they cost a shitload to design—hundreds of thousands of dollars just for a piece of wood. But everyone thought they were real.” In time, a specially set up subsidiary of Oracle called NCI (Network Computer, Inc.) managed to build about thirty thousand of the devices with the help of some hardware partners. One version, now called the NIC (New Internet Computer) struggles on today, mainly as a machine that Oracle donates to schools.2
Having gone so extremely public with his revolt from the whole client/server model, Ellison could hardly allow his engineers just to go on developing software as if nothing had happened. Ellison had spelled out his view that client/server was an evolutionary dead end and that a form of servercentric computing based on the open standards and protocols of the Internet was the future for Oracle. However, getting Oracle to change direction was not just a question of articulating a new strategy and expecting people to get on with it. The efforts of the entire company, from the developers building the applications to the sales and consulting teams who were selling and installing the software, were dedicated to making client/server a success. It was what customers expected and demanded. Whatever Ellison might be saying about the network computer and a new form of computing based on the Internet, client/server was, quite simply, the environment.
As for Ellison, it was as if he had needed the speech and the reaction to it to create a new set of circumstances, a new reality that would force him to lead the charge toward Internet computing, whatever that would turn out to be, despite the skepticism of many of his followers. Much later he said to me, “Once I’m finally certain of the right direction, I pick a fight, as I did with Gates. It helps me make my point, and it makes it impossible to do an about-face and go back. Once a course has been plotted, I sail a long way off and burn my boats. It’s win or die.”
How quick was Ellison to understand the significance of the Internet compared with others in the computing industry? There were undoubtedly others who saw the light even earlier, but he got there pretty fast. Farzad “Zod” Nazem, who worked at Oracle for ten years before leaving in early 1996 to become Yahoo!’s chief technology officer, remembers going out for meals with Ellison and “giving him kind of free tutorials on what you could do with this stuff.” Mark Jarvis recalls a meeting at Oracle in February 1995 to tell Ellison about some software that a development team had come up with to allow access to the database through a Web browser that had been christened the World Wide Web Interface Kit. “My first slide was ‘Here is the Internet, Larry.’ Larry was, like, ‘Next slide.’ Next thing, I pulled up the World Wide Web and Larry was, like, ‘Next slide.’ So Larry was already into this. He knew it all. I thought I was educating him, but . . . Larry is always three months ahead of everyone.”
What distinguished Ellison’s take on the Internet from that of many other enthusiasts was that he saw it not just as a great new way to get information and go shopping—lots of people quickly understood that once they saw the first Mosaic Web browser—but as a platform for a completely different and infinitely superior kind of computing. Right from the outset, he believed that the Internet really did change everything. Six years later he says, “Internet computing is the last architecture. There will be no new architecture for computing—not in a thousand years—not ever. I know it sounds a bit crazy. But with the Internet, computing has adopted the same architecture as all the other major networks—the hundred-year-old telephone and electric power networks and the two-thousand-year-old aqueduct network. Internet computing centralizes data storage in huge databases and processing on large servers, while distributing information on demand across a global network. Internet computing hides complexity, provides economies of scale, and delivers information faster.”
Softwar Page 6