% fortune -ae paul murphy

Oracle Solaris, Linux, and OpenSolaris

As the pennies start to fall in the Oracle/Sun merger we're seeing a lot of fear and ignorance getting expressed as corporate confusion, insensitivity, and general silliness - the company's websites, for example, now generally refer to Solaris as "Oracle Solaris" - naively falling in line with the IBM community press's use of "Sun Solaris" in an apparently deliberate effort to ghettoize its user community.

Meanwhile lots of other people are happily spreading FUD about OpenSolaris or disguising attacks on Linux as speculation that Oracle will downgrade its Linux support.

Although much of this is deplorable, most of it is also just par for the course, effectively collateral damage as the players in two big organizations merge interests, the PR folk get left behind, and the winners and losers in each organizational sub-group slowly get sorted out.

The most difficult and mission critical component of this merger will, I think, come with respect to sorting out the two support organizations. Both were pretty good, both had enormous internal complexities and dis-satisfactions arising mainly from the reality that support picks up the pieces every time sales ventures into new territory - through acquisitions, product expansions, or simply in response to customer pressures.

On the former Sun side, for example, identity management represented a customer necessity they simply couldn't say No to - just as customer demand for Lintel forced the former Oracle to periodically grab and freeze a Red Hat release for long enough to make applications support on it affordable for the customer.

The customer pressures driving these things won't change, but the new Oracle can offer better solutions at lower cost: shifting, to continue the example cited above, OS level Lintel support to the people who came from Sun, and recreating identity management as a unitary applications product that plugs into the appliance computing stack when needed.

This kind of organizational shuffle, in which responsibilities move to those closest to the mid point between customer and technology and the outliers get trimmed off, is a logical fit with Oracle's overall corporate strategy: sell what's selling to those who want to buy it, but direct business development dollars toward a revenue shift from licensing to support through an increasing focus on delivering highly reliable, functionally "deep", application services to customers.

Basically what this means is that if an Oracle customer wants to spend an extra few million shareholder dollars running Peoplesoft in client-server mode against DB2 under zOS, Oracle will take the money and do everything it can to help him make that happen - but the guy doing it has to be prepared to see his bosses reading Oracle ads offering to replace all that 70s stuff with a SPARC/Solaris based application appliance for less than five cents on the infrastructure dollar and no new application support cost.

It's this interleaving of long term strategic focus and short term customer response that places Linux and OpenSolaris in the merger crosshairs - not as things to cut, but as places meriting some additional senior management time and budget dollars.

On the Linux side this happens because Linux provides entry to the x86 market dominated by Microsoft. Tens of thousands of small to mid range businesses use Windows servers to run Windows applications - and every single one of those is a sales target for an Oracle guy armed with Lintel applications and a safe upgrade spiel guaranteed to deprive the Wintel clone driving that company's IT decisions of the forklift argument against looking at the Oracle offering.

Thus, the bottom line on Oracle's commitment to Linux is, I think, that it will gradually increase support for Linux until sometime after Microsoft switches to Unix for its business class server software.

Notice, however, that a commitment to Linux isn't a commitment to make or sell x86 hardware - on the contrary the only thing Oracle can possibly find attractive about this business is Sun's HPC position and relationship with AMD - assets that could leverage an interesting partnership arrangement with somebody like HP or Dell.

OpenSolaris is both different and the same. Most obviously, there are no significant organizational conflicts because Oracle's previous commitment to OpenSolaris was paper thin. Least obviously, but most importantly, OpenSolaris was a project with enormous promise that got decapitated when Sun posted a Linux guy at its head and went off to pursue its near death misadventure with x86 - a fate, and a direction, from which Oracle has now rescued it.

With that mistake soon to be safely in the past, the OpenSolaris group can stand on its own and can be expected to make the obvious decision that will give it immense strategic value to Oracle: stop trying to be a better Linux, and go be a better AIX instead.

The reason for that is both simple and complicated. On the simple side, OpenSolaris on x86 is a solution without a real problem, because there just isn't much you can do on x86 that you can't do with Linux - so sure containers, crossbow, zfs, dtrace, and all the rest of it are just as cool on x86 as they are on SPARC, but the limitations you run into between x86 hardware and the people you can get to make it work, mean that the real gains companies can get from using OpenSolaris instead of Linux on x86 are both too small and too unlikely to warrant the additional effort.

On the complicated side the fact is that IBM has nowhere to go but Linux - and if Oracle can help get OpenSolaris accepted as a valid alternative on IBM's hardware, it will open up a lot of accounts. Basically, where IBM account penetration is the goal, OpenSolaris is the obvious answer because a lot of IBM shops have somebody who can understand and champion what the technology offers: and once it's in the door - even if its only on x86 stuff - its value and availability on IBM's larger scale SMP hardware will create internal conflicts whose net effect will be to eventually break that account wide open.

So the bottom line on all this is simple: the forces that drove Sun to support Linux and open source Solaris haven't changed - and we can expect Oracle to sort itself out accordingly with solid support for applications on Lintel and for OpenSolaris and its applications on IBM's hardware.


Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.