What if Oracle bought Sun Microsystems?

If Oracle picks up where IBM left off, database, storage, and Java customers could be in for radical changes

So much for that. According to reports, Sun Microsystems has rejected IBM's buyout offer, apparently preferring to stick to the dogma that says Sun is still a vibrant, thriving company with a bright, independent future. At this juncture, I'd like to take a moment to offer the following public service announcement to any Sun employees in the audience: That's not Kool-Aid you're drinking.

More to the point, with IBM out of the picture, we're left with the uncomfortable prospect that some other suitor might step in to fill the vacuum. My pal Paul Venezia thinks Cisco might be interested in acquiring Sun, and one of the kookier rumors has Dell as a possibility.

[ For the full scoop on the IBM-Sun negotiations, see InfoWorld's special report ]

Neither scenario seems likely. Both assume that Sun's server business is its most attractive asset, which I question. If you added all of Sun's server revenue to Dell's, it still wouldn't equal what Hewlett-Packard earns from server sales. And if Sun were to simply disappear, Dell would surely gain at least some of Sun's share of the server market anyway -- so what would be the point of a buyout? What's more, I doubt that either Cisco or Dell would know what to do with Sun's software business.

But Oracle would. In fact, Oracle might stand to gain even more from Sun's software assets than IBM would -- so much so that I rank Oracle as the top (perhaps the only) potential buyer left for Sun. The shame of it is that if such a deal were to go through, I suspect that in the long run, Oracle's gain would be our loss.

Is MySQL doomed?
First off, let's dispense with the notion that Oracle would acquire Sun with the goal of scuppering MySQL. That wouldn't be worth Oracle's time, let alone the money. Oracle's annual database sales are measured in billions; any sales lost to MySQL barely make a dent.

And that even assumes that Oracle is losing sales to MySQL, which it isn't, really. Anyone who understands databases knows that for the high-end, mission-critical applications that drive Oracle's highly lucrative sales -- government, finance, oil and gas, and so on -- MySQL isn't even a player.

Sun knows it, too. When Sun acquired MySQL AB for $1 billion last year, CEO Jonathan Schwartz said that while Sun "will scale MySQL to extraordinary heights," it would not try to compete with Oracle. It was MySQL's rapidly growing installed base of 11 million deployments that attracted Sun, not its technology.

Who could be more interested in that installed base than Oracle? Although MySQL is a poor competitor, it would make an excellent complement to Oracle's existing database business. In fact, I'd say Oracle tipped its hand in 2005 when it acquired Innobase, makers of the leading transactional storage engine for MySQL.

MySQL's user base sits largely at what you might describe as the low end of the database market: Web sites, departmental servers, and single-use installations, for example. That's a niche that has been slipping through Oracle's fingers as its own database has grown in sophistication, complexity, and cost. By acquiring Sun, Oracle would be able to offer its customers a popular, well-recognized entry-level database, with an implicit upgrade path to Oracle's proprietary product as those customers' needs grow.

Oracle would be well served
Sun's other assets might seem like a poor fit for Oracle at first, but maybe not. One of the major complaints about a possible IBM/Sun merger was that it would leave the fate of Solaris up in the air. IBM already has AIX to worry about, and despite its heavy reliance on Linux, it has always preferred to partner with others rather than maintain a Linux distribution of its own. Oracle, on the other hand, already supports and manages a Linux distribution, and given the long history of Oracle databases on Solaris servers, it might actually see owning Solaris as an asset.

Among Solaris' most ballyhooed features is the ZFS filesystem -- which, according to the InfoWorld Test Center, "far surpasses anything available now on any platform" in terms of flexibility and scalability. Another is DTrace, a debugging feature that can be used to tune system performance. These kinds of features are just the things to appeal to performance-minded Oracle DBAs who want to manage their databases with minimal headaches.

Whether Oracle would be interested in continuing Sun's physical storage and server hardware businesses is another matter. Being the world's second-largest software company is probably enough, without entering into a whole second market. But let's not forget, Oracle has already released its own, branded hardware offerings with the help of HP. One could easily see it partnering with HP again if the numbers seemed sound, licensing Sun's technology for production on HP's manufacturing infrastructure. (In fact, one unsubstantiated rumor suggests that Oracle and HP have already proposed just that.)

So that gives us Oracle databases running on Oracle-branded servers and Oracle-branded storage hardware based on sophisticated Oracle storage technologies running on an Oracle OS. It sounds pretty good -- at least, if you're an Oracle sales rep. From a customer's perspective, though, I wonder.

Java, brought to you by Oracle
That brings us, finally, to the crown jewel of this deal for Oracle, which would be Java. Don't kid yourself: It's almost impossible to overestimate the importance of Java to Oracle. Java has become the backbone of Oracle's middleware strategy. And Oracle already owns BEA. What do you suppose it would be worth for it to become the leading provider of Java technology in the world, from the Java EE stack right on down to the core JVM itself?

Let's look at the numbers. IBM offered just shy of $7 billion for Sun. Oracle paid $8.5 billion for BEA, and that was after BEA rebuffed its initial offer of $6.7 billion. Mind you, Sun also rejected IBM's offer. But when Oracle wants something, Oracle gets it -- just ask PeopleSoft (which, by the way, set Oracle back $10.3 billion).

So let's recap. In this final scenario, we have enterprise customers running applications written for an Oracle-managed language platform, running on an Oracle-branded application server, which communicates with an Oracle database on Oracle-branded servers that talk to Oracle-branded storage hardware running an Oracle filesystem on an Oracle OS.

Oh yeah, and Oracle would own MySQL.

Of course there are a few stumbling blocks. Maybe Sun's aggressive efforts to open-source its entire software portfolio have made many of its assets seem like nonstarters for a company as focused on proprietary software as Oracle. Maybe Oracle wouldn't want to bother with licensing JVMs for mobile phones, Blu-ray players, and other consumer technology devices, which supposedly accounts for much of Sun's Java revenue. Maybe this is the wrong time for Oracle to get distracted from its ongoing efforts to consolidate its enterprise applications.

Or maybe, just maybe, we all might live to regret Sun turning down that offer from IBM.

Copyright © 2009 IDG Communications, Inc.