Maybe. Maybe not. I've had nothing but bad experiences with Oracle in several jobs now and will emphatically advise against it at every opportunity. Postgres may not be a drop-in replacement for every Oracle use case but for those that are I'll recommend it first any day of the week. Oracle in a job description is a big red flag for me.
Because a properly installed and maintained Oracle is pretty reliable and straightforward to use. If you even can run your application on Postgres then you wasted your money buying Oracle; you aren't using it for what it's for[1].
[1] I am aware that convincing people that they might need to in future is key to Oracle's business model, but that's beside the point.
Technically. In two different jobs now I've seen a small army of consultants unable to keep an Oracle cluster running as reliably as a single Postgres instance, despite jaw-dropping license fees. It's also pretty painful to use, from simple command-line interactions to dumping & restoring data to interfacing with any code that isn't "enterprise". Blob performance was also astonishingly bad.
I guess you're stuck with it if you need to run Oracle financials or Peoplesoft or something like that but one of my professional goals is to never work for another company big or dumb enough to use either.
That says more about consultants than it does about Oracle ;-)
Oracle (the database) in the hands of anyone basically competent is easily capable of 5-9s reliability. Not that I'm saying Postgres isn't mind. I'm just saying, use the right tool for the job (and hire the right people).
I find DataPump very easy to use, and very fast, it's I/O bound even on serious storage arrays. YMMV. SQL*Plus is definitely showing its age I agree.
Most obviously, if you want to run one of Oracle's layered applications (e.g supply chain management or whatever). Sure you probably could beat it into running on Postgres but why would you?
Oracle (the database) is meant to be the integration point for Oracle (everything else). If you aren't running any of Oracle's other products, then it boils down to a pure technical decision - do you need feature X? Is it cheaper to buy it or build it? For example do you use OpenFiler or buy VVM or use ASM that comes with Oracle since you need a database anyway? Or maybe you have exotic spatial data needs. Or you want virtual private database/label security (http://wiki.postgresql.org/wiki/RLS - Oracle has had this for years, the Postgres guys are still talking about it, the MySQL camp don't even know that there is such a thing).
Now, I am a long-time Oracle user, and even I say, unless you know and can say specifically why you need Oracle, then you should use Postgres or SAP/DB (or in place of Oracle XE, maybe even SQLite). But if you do need it, then it's Oracle or DB/2, choose your poison.
Well, despite some promising open-source efforts in the area (PostGIS being most prominent) Oracle is still tops when it comes to handling geo-spatial data. Unfortunate, but that's the way it is.
Oracle is like the C++ of databases; Postgres is like the C#. Oracle is ugly and complex, but it can be made to do amazing things that no other RDBMS can do. Postgres is really fast out of the box for 90% of most use-cases, but it still can't do the things Oracle can.
I was actually thinking of exactly this analogy earlier. It holds also in that a lot of people use C++ because they think they need it but they could in fact use something simpler.
