Hey, do yourself a favor and pick up some good books on database schema design and things like foreign key constraints, triggers. It'll do you a world of good.
Database design is as much art as science, and as much black magic as anything else. Each RDBMS system will have it's own quirks, and you'll likely find each one is 10 to 50 times faster after proper tuning than before, so it's hard to pick the right database by just installing it and doing a quick benchmark. The database that seems the slowest at first may wind up tuning up to be faster than all the rest.
Lastly, remember that for Oracle to do the job, you're gonna need an expensive license, and that license will likely be tied to how fast your machine is and how many users you have.
You may find that setting up a replicated Postgresql or MySQL database farm or moving your data into ISAM files or something on a couple of real fast boxes can be justified just by the much lower licensing cost.
I.e. if your Oracle license is gonna be $250,000 US, then it might be cheaper to spend some small portion of that on a couple of Dual Athlon MP systems with gigs of ram and big giant arrays of disks hooked up via Fibre optic or multiple Ultra 160 SCSI cards.
By the time you spend as much on hardware alone as your Oracle license would cost, you're gonna have some pretty massive machinery for running Postgresql or MySQL.
It sounds like you're building a huge data store with basically static data. In this case both database tuning and hardware / OS tuning are gonna be critical to getting good performance, so whichever databases you want to try, spend some time getting both the box they will run on and the database themselves tuned up.
