Garry pointed to IBMs recent release of a freebie version of DB2 as being a solid indication of a future trend in which the big software players will move away from software license revenue as a business model. The freebie database in question, DB2 Universal Database Express-C (Community Edition), will have no limit on the number of users and no limit on database size, being deployable on up to two processing cores and up to two dual-core chips on x86 systems. The maximum memory size is 4GB. That generous dual-core spread and lack of limits is in stark contrast to Sybase Express, SQL Server Express or Oracle Express, which in their limitations are designed to be little more than development platforms.A real, functioning, production-level database thats free is an indication of the "further erosion of the concept of software licensing," Garry said. "Ten years from now, people arent going to buy software at all. Theyre going to go to Oracle or IBM or CA or whoever, and theyre going to hire them for their expertise in something. Companies will be disaggregating their business. " So its not necessarily having machines or having somebody run the whole operation," he said. "Its sending people in to look at the business, to design something specifically for your business, so [the business itself is] not doing it. In that world, its an ASP model, its not something were buying." But even if the shift to the ASP model is in the offing, it wont come anytime soon, Garry predicted. The revenue hit would be too hard, and Wall Street "would go bananas," he said. But Oracle in the meantime could use the subscription model to upsell, offering support and other options for a fee to those who want it. That would drive adoption by people who download products such as JBoss or MySQL but never pay, while also attracting the Global 5000, who may develop with free versions, but who typically want not only support but also indemnification when they deploy, Garry said. "Theyre going to want support, and not just support but indemnificationsomebody like IBM standing behind the product," Garry said. The margins for support are huge. Oracle scores some $3 billion a year in new license sales, and they made about $10 billion last year, most of it from support. But what does going to an ASP model have to do with purchasing open-source companies? Couldnt Oracle just move its own applications to the model? Garry suggested that open source beckons as a business Oracle just cant ignoreone where components are relatively cheap to buy, and by buying those components, a significant amount of users come with them. Oracle said in its release about the Sleepycat buy that Berkeley DB is said to be the most widely used open-source database, for example, with an estimated 200 million deployments. At the same time, the market opportunity presented by the embedded database space is expected to grow from $2 billion in revenue in 2005 to $3.2 billion in 2009, according to IDC data cited in Oracles statement. As it is, Oracles embedded database product line also includes Oracle Lite for mobile devices and Oracle TimesTen, designed for high-performance, in-memory database applications. "Im assuming [Oracles] hope is that by buying these [open-source components], the switching costs will be too high, that people will stay with it," Garry said. Thus, Oracle controls the space and hopefully upsells. Even if only a tiny percentage, say 1 percent, of open-source users eventually buy from Oracle, it will still be a useful signal to Oracles core audiencei.e., the enterprise software buyerthat open source is legitimate and has large companies such as Oracle solidly behind it, so deploying it more widely in the enterprise is a safe bet. At any rate, Oracle is smart enough to know that such a scenario will eventually happen whether or not Oracle controls the adoption of open source. Why not control that adoption if it canand the purchase of Sleepycat and InnoDB clearly shows its canand make money off it? "[Oracle is likely saying], Wed rather control our future rather than have somebody dictate to us," Garry said. With regard to how JBoss would benefit Oracle, should that acquisition rumor prove true: It would allow the company to get a handle on its many disparate application suites, according to Judith Hurwitz, president of Hurwitz & Associates. "The Fusion [Middleware] spec is not codified softwarethat means they need middleware now. They cant wait," said Hurwitz, in Waltham, Mass. "Theyll use JBoss as integrating middleware ... [to develop] a consistent set of services across all their suites. Just to maintain each separate [infrastructure] is very expensive." Fusion, according to Hurwitz, is "just not done. Its not consistent across the entire stack." Editors Note: This story was updated to include comments from Judith Hurwitz. Check out eWEEK.coms for the latest database news, reviews and analysis.
"[DB2 UDB Express-C] is not just a development platform," Garry said. "You can actually deploy real production applications on it."