It seems to me that one of the things that came clear through last week's Oracle openworld conference is the extent to which the company is preparing itself for directional change with the fusion applications.
Originally Oracle was a company dedicated to making and licensing data management software. Then, basically since 7.3, it became a company almost wholly devoted to selling almost anything it could hang a company label on - and now, I think, it's about to transition to a new focus as a compelling business applications developer.
The fusion applications were originally intended to reduce the costs and complexities of supporting about 37,000 enterprise customers by integrating the lessons learnt from Oracle's own enterprise suite, PeopleSoft, Siebel, and the J.D. Edwards stuff into one consistent package. Now, however, I think they're intended to save the company - because it's simply not going to be possible to sell most of the database, management, and BI products once forthcoming MySQL and Postgres benchmarks running on cheap Solaris/SPARC gear have been transformed, through a year or two of actual customer use, into industry standard expectations.
In other words, think three years out and by present trends an open source database running on SPARC/CMT can be expected to outperform a grid based proprietory system by a factor of ten on throughput and two on cost - meaning that customers choosing between licensing something like Oracle's database management software and mySQL will have compelling incentives to go with the open source solution.
To a company like Oracle that means the handwriting's on the wall with respect to RDBMS and related revenues -and it seems to me that the strategic option they've chosen in response has been to move up the food chain by concentrating on producing a complete applications stack. That works because enterprise application customers care about the total cost of using the application - not whether Oracle internally uses some of that money to support an inefficient database/hardware paradigm.
I hope that's right because my bottom line on this is that it could be great: I've long been a big fan of both Oracle's enterprise applications and Peoplesoft's and so think that if fusion combines and advances both while leaving more of the general infrastructure market to open source, it could work out well for everybody.