Softwar

by Matthew Symonds

  Not only did Oracle appear to have the right products for the times, Ellison’s unrelenting focus on the Internet and the e-business revolution had also turned out to be spectacularly right. The latest version of the flagship database, Oracle 8i (“i” as in “Internet,” of course), had shipped in March 1999, and Oracle’s first true Web-based ERP applications suite, 11.0, was actually winning market share against SAP’s product because, with “Y2K” fast approaching, it was proving much quicker and easier to install. According to Dataquest, Oracle was the only leading ERP player to show significant growth in the first half of 1999. While PeopleSoft suffered a 53 percent revenue decline and SAP managed only a measly 3.8 percent year-on-year growth, Oracle’s applications revenue grew by nearly 30 percent. As Ellison said, 11.0 was “really smoking” the competition. As for Oracle 8i, when announcing its forthcoming release at a New York press conference in September 1998, Ellison had declared that 8i was not just a database but a platform for Internet computing. With its built-in development tools, it was all that companies needed to deploy business applications that could be accessed using a standard Web browser.

  The “first Internet database,” as Ellison described it, was packed with hip-sounding new features: a built-in Java virtual machine that could execute applications written in the new Java programming language, which was taking the development community by storm; the latest messaging technology and directory services; additional Common Object Request Broker Architecture (CORBA) interfaces and better support for clusters. Most customers would initially buy it for the traditional reasons—increased speed and performance—but others raved about 8i’s Java capabilities from the outset. One was Jeff Grant, the IT manager for the recording company Nettwerk Productions, which represents Barenaked Ladies, Gob, and Sarah McLachlan. With sixty thousand subscribers to its mailing lists, he described the Java Mail package as a “lifesaver”: “No more using another mail server, no more screwing with building database cartridges for the server. Now you have the ability to call the Java routine straight from the database and no longer have to use a go-between.” Similarly, Grant found that if he wanted to make a change to any of the applications he was running, instead of having to upgrade the software on five thousand PCs, he could now make the changes in one place. “Developing apps like this makes all the sense in the world,” he said.

  Jeremy Burton, then responsible for server marketing at Oracle, said, “The features that are important now are scalability and reliability. But over time customers will figure out what this Internet model can do for them.” Merv Adrian, an analyst at Giga Information Group, observed, “You can’t just sell a database by saying it works better. It has to have the latest thing, like new, improved XML support with JavaBeans. 8i represented a dramatic shift in focus to Internet-driven changes. People are buying it because it meets today’s and tomorrow’s needs.”

  One of the most significant 8i innovations wasn’t ready until more than a year after the database came out. Called Internet File System, it allowed any Windows application file, such as Word or Excel, or any Web page or e-mail, to be dragged and dropped into a directory and integrated into the database. Among other benefits, developers would be able to build Web applications that linked both relational data and Windows files. When Ellison said that 8i was not just a database, he meant that users would no longer have to rely on Windows. The database was agnostic about which operating system it ran on! They would be “insulated” from the operating system; in other words, the OS would be merely a utility.

  It was an important propaganda message to get out at the same time Microsoft was launching the supposedly much improved Version 7 of its SQL Server database. Unfortunately it lacked a reliable operating system to run on, and Microsoft was still laboring mightily to finish the long-overdue new version of Windows that it believed would finally win SQL Server 7 a ticket into the corporate data center and thus into Oracle’s heartland. Ellison was saying, in effect, that for all Microsoft’s efforts to become an enterprise computing player—it claimed that Windows 2000 was one of the biggest and most difficult engineering projects of all time, with armies of developers struggling to find the bugs in the 45 million lines of code—it was already out of date. To amplify the message that operating systems were not very important anymore, 8i would also run on Linux, the free operating system developed by the open-source movement. Furthermore, Oracle, in partnership with hardware companies such as Sun and Hewlett-Packard, would market something Ellison called a “database appliance.” Known as Raw Iron, it would run straight out of the box using a stripped-down operating system. Raw Iron didn’t take off commercially, but it reinforced the idea of simplicity versus complexity.

  If Ellison was happy to describe Oracle 8i as the Internet database, he had no difficulty officially christening the forthcoming application Release 11i the E-Business Suite. For Ellison, 11i would be the summation of everything he had learned about the applications business since his forced introduction to it in early 1998. Above all, it would deal conclusively with the curse of fragmented data that had begun with client/server and that best-of-breed had exacerbated regardless of whether applications were Web-enabled or not. Apart from its new Internet computing architecture, which superseded that of 10.7 NCA and 11.0, the new suite would have three unique characteristics to ensure that Ellison achieved his holy grail of “all of your information, all of the time in just one place.”

  For the first time, 11i would offer tight integration between traditional enterprise resource applications that managed back-office functions such as financials, HR, and manufacturing automation, with the newer customer relationship software that ran marketing, sales, and service. In addition, 11i included an advanced planning suite, enterprise asset management, and online procurement. It is axiomatic that information about any customer transactions, supplier transactions, or internal business processes can affect every part of the organization. The best example of this is the apparently simple one of orders. From sales to marketing to customer service to supply chain to production, just about everybody needs to be able to access order information to do his or her job properly. Without integration between front- and back-office software, the ideal of “order information everywhere” is unachievable.

  The second claimed point of difference for 11i was its completeness. Ellison had realized that data fragmentation was not the only reason for the difficulty users of enterprise applications had in getting worthwhile information about their businesses out of their multimillion-dollar systems. Another was the problem of gaps in automation. If a business flow was interrupted by manual processes at any stage, it almost always meant that the information stored was, at best, partial. By attempting to provide all the core functionality to run most businesses from end to end, Oracle was promising seamlessness and with it the speed, efficiency, and responsiveness that were supposed to be inherent in the idea of e-business.

  The key to everything was the seemingly esoteric concept of a common data model uniting every piece of the suite. Every module—and there were about 140 of them—would be written to the same shared data schema, allowing semantic consistency (for example, the definition of a customer remained the same no matter from which application the information was coming and could thus be shared by all the other applications in the suite) as well as a complete view into every transaction. The underlying simplicity of the architecture was designed not only to make the gathering of useful business intelligence much easier than with multiple component systems but also to make implementation radically easier and therefore cheaper and quicker.

  The scale of Ellison’s ambition with 11i was breathtaking. While Microsoft boasted about the size of the project to build Windows 2000, it was, at the end of the day, only an operating system. The E-Business Suite, by comparison, had modules designed to run nearly every aspect of even the biggest and most complex multinational company’s operations. From its inception, more than four thousand developers had been working on it nonstop
(and still are refining the software and adding features demanded by customers or required to match competitors). When it shipped at the end of May 2000, as well as a new architecture, at least 40 percent of the modules themselves were entirely new. When Ellison described it as the largest and richest software product ever created, for once nobody accused him of exaggerating. The question was: Could Oracle pull it off?

  Although Oracle first started talking about 11i in April 1999 at the Oracle Applications Users Group’s (OAUG) conference in San Diego, the official announcement of a shipping date and what customers could expect from the new suite had to wait another five months. When ten thousand Oracle users showed up at the fall OAUG conference in Orlando to see Ellison make his keynote speech in the flesh, all they got was his image beamed from Redwood Shores on four giant screens. Many of them were also disappointed by what he said.

  After an almost perfunctory fifteen-minute ramble, throughout which he appeared to be wrestling with an uncomfortable earpiece, he declared that the anticipated November launch date had been put back until February 2000. The suite would be complete only when the CRM component became available sometime in the second quarter. Ellison explained that technology glitches were causing the delays: he had also decided to include a brand-new order management module that had been scheduled for a later release, rather than bolt on the old one as originally planned. One man in the audience, apparently speaking for a good few others, groused that the constantly shifting release dates left him losing credibility with customers, who were forced to put updating their applications on permanent hold.1

  Both Ellison’s apparently casual remark about the decision to go with the new order management module and the pressure from customers eager to get their hands on 11i would have profound consequences. But the immediate aftermath from Orlando, where Oracle executives had been energetically briefing analysts about the wonders of the new suite, was pretty positive. Gartner Dataquest wrote bullishly that 11i, by making Oracle the first big applications vendor to “use ERP/CRM integration as a loud, primary marketing message,” would only add to the superior momentum of Oracle’s license revenue growth over its rivals. It noted that apart from being easier to manage and deploy, the suite approach, as defined by Oracle, would provide “richer business intelligence and decision support.” It summed up: “Dataquest awards Oracle a thumbs-up for pouncing on an opportunity to present a clear message. . . . Oracle would benefit wildly if ERP/CRM integration becomes the key driver for vendor choice in the enterprise applications space.”

  There was, however, a word of warning—“Of course, Oracle must now deliver”—which the author had some doubts about, given “its somewhat less-than-sterling past performances” in applications. The META Group made a similarly qualified prognostication: “Oracle, somewhat hampered in the past by offerings that lacked coherence and differentiation, has recently demonstrated a consistent vision. Although it remains to be seen how smoothly its plans are carried out, Oracle is positioning itself to aggressively increase its share of the overall enterprise application market.” The analysts were saying that for pretty much the first time Oracle had a potentially appealing story for customers who were sick of bearing the risks and costs of integration. Their caveat was that Ellison’s “change the game” strategy required near-flawless execution. Oracle was on a roll, but if it became apparent that it had bitten off more than it could chew, the company could expect little mercy.

  Many of the industry analysts regarded Oracle as an arrogant company that had never shown much interest in engaging them in dialogue. But it was more complicated than that. Oracle had spotted what it thought was a serious conflict of interest for the tech consultancies when it came to accepting the single-data-model concept behind the suite. Mark Jarvis, Oracle’s head of marketing, says that he never expected an easy ride: “I thought that the Gartners and the Forresters of this world would find it hard to accept the idea of buying everything from one vendor. If they no longer have to advise clients that you need to buy this software for marketing and this software for sales, ending up with buying software from fourteen different vendors, they will be out of business. Their advice and their expertise are no longer important. You also have to bear in mind that some of those firms are in the systems integration business too, so we were threatening their revenues on two fronts. The financial analysts were likely to be more open-minded. But many of them were advising people to buy the likes of Ariba and Commerce One [which were threatened by Oracle’s suite strategy], so they had some vested interest too.”

  However, the initially largely favorable analyst coverage, which Oracle suspected wouldn’t last, added to the pressure on Ellison to ship 11i. There was pressure of a different kind from Wall Street. Because customers sit on their wallets when they know a radical new release is imminent that is more than just an upgrade from existing software, earnings can stall. With Y2K out of the way and company IT spending on e-business-enabling software expected to boom, for Ellison the cost in revenue postponed or even lost because of further delays in 11i was unacceptable. If that meant putting heat on development, especially Ron Wohl and Mark Barrenechea, to declare the software customer-ready, then so be it.

  As Ellison frequently says about himself, he doesn’t have any difficulty in making decisions, even if they sometimes turn out to have been wrong. In this situation, he concluded that the risk was asymmetric. As well as the hit to Oracle’s enviable earnings momentum, with its inevitable impact on the company’s heady stock price, if Oracle were seen to be struggling, critics would be only too happy to declare Ellison a victim of his own vaunting ambition. At the same time, more delays would give rivals time to undermine, perhaps critically, the credibility of the suite concept before customers had even had a chance to try it. On the other hand, pushing 11i out of the door before it was finished ran the risk of generating a raft of unhappy customer stories for the analysts to feed on.

  But maybe that wouldn’t happen. The certainty of a launch date would undoubtedly concentrate minds. Maybe 11i would work perfectly out of the box. In the software industry, it was not unknown for miracles to happen. Even if a miracle didn’t happen on this occasion, Ellison was cynical enough about the development process to believe that however much work is done in-house to iron out problems, it’s not until the product is in the hands of real customers that you really start to find out what’s right and what’s wrong. Ellison was also prepared to take the apparently sensible path of having a controlled customer release and staggering the launch, rolling out first the more mature ERP part of the suite in February and then adding the CRM modules and the brand-new order management application in May.

  It sounded prudent enough, but the reality was very different. In the first place, Oracle was breaking one of the cardinal rules of the applications business: don’t release software to customers until you’ve tried running on it yourself. While much of 11i was based on the custom-built applications that Oracle had used to turn itself into an e-business and save $1 billion, Oracle would not have time to move its own operations onto 11i before it began selling it to customers. They, rather than Oracle, would be playing the guinea pig.2

  To make matters worse, the idea of a controlled release to early-adopter customers who were ready to trade some pain for the gain of getting their hands on the software before competitors wasn’t strictly adhered to. Part of the problem, according to Ellison, was that customers in Europe were getting their systems ready for the introduction of the euro and had been planning on implementing 11i as part of that process. If 11i came out later, they would not only lose vital time but would incur heavy costs from extending their contracts with systems integrations firms. Ellison says, “I had already delayed the 11i release date, and that forced our customers to revise their implementation plans. There was tremendous pressure not to delay again. Our customers had their resources all lined up and were ready to install our software as soon as we released it. It was a bit like the beginning of World War I:
once the troop mobilizations began, there was no turning back.”

  Under the circumstances, a prudent general might have done everything possible to eliminate any unnecessary risk. Instead, Ellison decided on the software equivalent of a blind cavalry charge. It had never been the intention to put the new order management module into the early releases of 11i. The idea had been to start off with the old one and upgrade customers later. But because the launch of the whole suite had been delayed, somebody suggested that it might be possible to sneak the new and much more capable system in at the last moment.

  Ellison says, “We had just finished a complete rewrite of our order management system. It was a huge job. The new order management system was dramatically better than the old one, but it hadn’t been tested yet, so we thought it was much too risky to put it into the suite. Then someone—not me this time—had this very innovative idea: let’s put both the old one and the new one into the suite. That way, conservative customers could stick with the old reliable order management system while our most avant-garde customers could get the benefits of the new one. Cool idea. Unfortunately, it turned out to be technically impossible, but it was a hell of an idea, and Ron and I took our time before we gave up on it. There were five stages to the order management decision. Stage one: We have time to put the new order system into the suite; let’s do it. Stage two: No, too risky; we have to go with the old order system. Stage three: Let’s put them both in; absolutely fantastic idea. Stage four: Fantastic idea, but we can’t make it work. Stage five: Well, then, let’s put in the new one. It was my decision. It’s in my nature to go for the highest-risk/highest-reward option. That’s my cross to bear. We went from having an average order management product to having the premier order management product on the planet. The only problem was that it was brand new and we hadn’t done enough testing.”