A thought struck me this morning: would you really want your database/application server/office productivity suite/etc. developer to also be the author of your operating system? On one level, the answer is an easy 'yes': tighter integration between the OS and the applications that run on it is a good thing.
On the other side of this coin, however, is the reality that today's integration is tomorrow's barrier to entry against all other applications. Take Microsoft, for example.
Microsoft has arguably done a very good job of encouraging third-party applications on its Windows platforms. But Microsoft has done less well once it starts to compete in a given application market against its partners. Even where the company has a financial incentive to boost the partner, it has a competing incentive to boost itself.
Of course, Microsoft is hardly alone in this. Oracle, IBM and others all face the same challenge: how to cooperate when you've grown so big and diverse in your product offerings that you can't help but compete?
Back to Linux/the OS question. Some Oracle customers are likely happy to have Oracle serve as Linux and application/database vendor to them. They like the integrated approach, just as many plug into the Microsoft ecosystem as a way to improve integration and (hopefully) reduce complexity. And some of us hope that Red Hat expands beyond its OS roots to move higher up the software stack.
But should we? One of the great things about Linux is that no company owns it. It's a common platform that is neutral in terms of the applications that it supports. Linus Torvalds is not making tweaks here and there to privilege his friends at SAP. He's working with the kernel development community to create the best OS, period.
Isn't this what we want? Aren't we better off having the OS vendor separate from the application vendor separate from the database vendor? Do we really want our consolidating, integrating industry to leave us with fewer choices, not more?