Thanks to everyone for all the feedback and ideas on this thread.  I asked 
another group the same question and I am learning that PostgreSQL would 
probably be easier than MySQL to port to.  It will take me a while to 
learn what all the issues are with the Oracle database.  Given the 
complexity of transitioning to PostreSQL or MySQL, it probably won't 
happen for awhile.  It seems that we don't have to pay for it right now:

http://www1.umn.edu/adcs/site/list.html

The goal of these university discounts is to get students addicted to 
programs that they will then want to use in industry work.  So money isn't 
the issue right now, but it is possible for the pricing scheme to change, 
though it probably will never cost enough to really matter.  The more 
important issue is what our students and staff are learning and how easily 
we can replace workers who leave the project with people who understand 
the software.  FOSS solutions tend to be more widely used because they are 
readily accessible to anyone and this can mean a better applicant pool. 
Such solutions also offer more control to the user because the code is 
available.  (It is sometimes too much work to edit the code, but you can 
at least know that you can stick with an earlier version and not be forced 
to upgrade to something you don't like.)

In academia, when we have a good idea, we often like to share it.  If we 
produce something great that uses proprietary software, like Oracle, it 
isn't as helpful as when it uses a FOSS package -- we can't share all of 
the code because some of it is proprietary.  The more we can work within 
the FOSS codebase, the better for the advancement of science.

Mike