Is Oracle doomed? Can it exist as an independent company beyond another 3 to 4 years? I doubt it. Have I got your attention yet? I'm sure all the Microsoft bashers have already revved up their flame mail. Don't worry, you'll get your chance, but keep reading before you decide I'm the craziest person in the world for predicting Oracle's inevitable demise.
Last week at the SAP SAPPHIRE '99 Users Conference, Microsoft announced a new world record for the SAP Retail Module benchmark. Microsoft and SAP jointly produced a benchmark number of 3.01 million sales data line items per hour, which is the best result ever published on ANY platform. Sun Microsystems held the previous record of 2.4 million line items per hour using a 32 processor UE10000. The complete server configuration tested by Microsoft (running on Compaq boxes) cost slightly more than $2 million or roughly half the price of the Sun configuration.
Let's jump to another topic for a second. How much would you pay for a toaster? That's right, just a plain toaster. Don't like toast? How about 27-inch TV set? Not fond of the boob tube? How about 100 frozen pork belly futures on the Chicago Mercantile Exchange? For the most part, all of these things are commodities. Each group's features are essentially the same and price has a huge part in determining which product you'll buy. Now let's jump back to databases.
Databases, more and more, are becoming commodities. Vendor A announces a new release to leap frog the competitors who are already planning to announce new state-of-the-art releases of their own. Features are nice, but everyone's got them. I'm not suggesting that Oracle, DB2, SQL Server, and the rest have identical features, but you have to admit that for the most part, a database is just a database. Features are important, but other factors drive what database you want and how much you're willing to pay for it. I think that most people consider performance and reliability to be the two most important price-setting features.
Have you ever seen one of those middle-aged guys with a $100,000 exotic sports car that he drives to the office and the corner grocery store to buy milk? Oracle has sold a lot of database products using a similar technique by playing to a customer's ego and fear. Here's the pitch, "Sure, our database is a lot more expensive, but do you really want to risk your business, or your job, by saving a few bucks? You might not need a 64-node processor today for your midsize departmental server, but what if you need it next year? SQL Server might be less expensive and easier to manage, but it can't scale. Don't worry, no one ever got fired for buying Oracle." (Please, no hate mail on this topic; Microsoft is admittedly a master of the "fear, uncertainty, and doubt" sale as well!)
Over the years, Microsoft has been most successful pitching SQL Server to cost-conscious customers who were willing to save a few bucks but sacrifice some potential scalability. Historically, SQL Server press releases have crowed about price and performance and downplayed overall performance. Oracle has kept its market share because Microsoft simply couldn't compete at the upper end of the performance scales. Now let's jump back to the SAP benchmark.
I could be wrong, but I'm pretty sure this SAP retail test is the first major, enterprise-class benchmark announced by Microsoft where SQL Server had the best performance on ANY platform regardless of system price. Who says SQL Server and Windows NT can't scale? By the way, SQL Server was faster AND close to half the price of the Oracle solution. Can Oracle out perform SQL Server in some cases? Absolutely, but the writing is on the wall. In the not so distant future, SQL Server and NT will be as fast or faster than any product Oracle can throw at it. When that happens, SQL Server and Oracle are essentially frozen pork belly futures or some other price-sensitive commodity, and most customers won't be willing to pay Oracle price premiums when SQL Server and NT combinations offer the same or better levels of performance. Can't Oracle lower prices so it's more competitive with SQL Server? Heck no. The core database engine is too big a slice of the Oracle revenue pie. Oracle can't cut database prices to compete against SQL Server without gutting its business model, profit margins, and R&D investments.
Much to the chagrin of my Microsoft buddies, I'll admit that Oracle has great technology, and I don't think that Oracle is going away. I just don't see how it can continue to exist as an independent company as Microsoft continues to narrow the overall system performance gap. But, I do think Oracle will have to merge with someone over the next few years. Lots of circumstances can change in Internet time, so several years is an eternity, but today the most likely combination would be some sort of Oracle/AOL/Sun alliance. Of course, this change won't happen overnight. Microsoft has years of necessary and difficult work ahead of it to make SQL Server and Windows 2000 Datacenter Server (Datacenter) a true 24x7, 99.9999 percent enterprise solution, but the little Dutch boy has his finger in the dike and there's no holding back the flood.