Microsoft's DNA won't permit Oracle-Sun deal

Ballmer knows his knitting

Comment When Steve Ballmer tackled the inevitable question on a Microsoft hardware acquisition, in the wake or Oracle's planned purchase of Sun Microsystems, he was "sticking to the knitting."

"I have no idea why a software company would buy a hardware company," Microsoft's chief executive is reported to have said.

That's not to say Microsoft is impervious to one big trend working its way through the market for storing, serving, and understanding information that Oracle occupies: the trend for getting fast access to huge quantities of data on massive networks and making sense of it. Far from it.

Increasingly, a hardware and software stack tuned to the needs of a combined database, storage system, and processing unit - an appliance - is seen as the way to tackle this.

Oracle last year announced the HP Oracle Exadata Storage Server and HP Oracle Database Machine, a box from Hewlett-Packard featuring a stack of pre-configured Exadata Storage Servers all running Oracle's database and its Enterprise Linux.

It's possible that Oracle's purchase of Sun could help it turn out more such boxes. The only question is why Oracle would want to bother taking on the burden of running a global, hardware design and manufacturing operation when its core competence is software.

This could be the ultimate act of hubris. Oracle's done well buying, assimilating, and monetizing software companies, so why not now try hardware? Or it could be business savvy: taking on a loss maker with potential to turn it around and sell it on, corporate-raider style.

If Oracle doesn't sell Sun's hardware business then it could well have decided the cost and expense of owning a hardware operation can be off-set by the value of building and tuning a combined hardware and software stack guaranteed to run. Something that is not hostage to the little unknowns of running its information software on a third-party's hardware. In this model, Oracle becomes the Apple of information appliances.

Microsoft is also addressing the appliance trend, but it's using a proven model that it's familiar with and that has been successful in the PC and server markets: building the software components and partnering with those who know hardware to bring its software to life.

The company's working with partners Bull, Dell, EMC, Unisys, and - yes - HP on server appliances. These are big companies that span mainframe- or Unix-class computing or storage power, with appliances due in 2010. The appliances will be based on a subset of technology in the version of its next SQL Server database, which is codenamed Kilimanjaro.

Overall, Kilimanjaro's focus is business intelligence, with "self-service reporting" and "self-service analysis." The appliance will provide large-scale detailed analysis.

Microsoft is no stranger to hardware and - as ever in weeks such as these - some will be holding up the mirror to others in the industry.

There's certainly been plenty of speculation over when Microsoft will make it's own mobile handset, speculation driven by the feeling that Microsoft must do something similar to the iPhone in order to answer Apple with Windows Mobile. And certainly, Microsoft has done hardware before - there's the Xbox, Zune, and various mice and keyboards.

Microsoft has also often been linked to HP, traditionally the company's biggest partner on Windows and in related consulting services.

But when it comes to hardware, Microsoft has always stuck to the software knitting and engaged selectively and in ways it can manage. Microsoft's approach on hardware is to partner, build an ecosystem, and then help tune and certify partners' resulting machines to its software.

Acquisition of a hardware company would break the DNA sequence and fundamentally change Microsoft in the way that owning Sun's hardware business will change Oracle. ®