When I worked at a software startup, we used firebird & db2 as light weight, client-side databases. Our network appliances ran MSSQL, which was fine. However, when we needed a real database, to hold real amounts of data, we used postgresql. Mysql can’t touch it for speed, features, and compliance.

However, given how often I use sqlite – amarok, firefox, etc – I had to vote for that as my favorite, in that it does more stuff I care about than any other database (since I don’t know what db most of the sites I like use, I can’t vote based on that).

At work, I help to develop a product with Oracle (and DB2) as a backend. In order to develop it and test it, I need to maintain several Oracle instances personally. I can attest that Oracle is byzantine and arcane. However, most of the reason for this is it’s ungodly number of features and level of flexibility. Oracle can be used to maintain and organize huge quantities of data.

At home however, I prefer PostgreSQL. It’s much simpler and lighter than Oracle (not to mention cheaper :P) It’s much easier and more intuitive for me to maintain. I generally avoid MySQL, because I actually find it more arcane than Oracle, while having fewer features than Postrgres and barely even keeping the Relational Model.

@naum: The relational model is by no means passe. The object-hash table structure has been around forever and BerkleyDB has been available since the 80s. The relational model is still in wide use. The is completely ignoring the rash of object dbs that popped up almost a decade ago and still provide headaches for developers today (mainly in the form of poorly designed ORMs.) The relational model can guarantee things like consistency and prevent things like redundancy. The relational structure provides information to the DB engine that allows it to more efficiently process your data. Object systems require one to code the data access algorithms oneself.

Document DBs do offer more flexibility than relational systems. The chief advantage of CouchDB for example is that it provide some structure to the data it stores, while sill allowing for that structure to change. This kind of flexibility is very important in agile systems, and can be useful for applications like say, a website, but I wouldn’t use it to store financial records.

I chose MySQL but I do most of my work with SQLite. I’ve also been looking into schemaless databases like QDBM and Tokyo Cabinet (or tokyo tyrant)

SQLite. Simple, compact, fast, ACID compliant, though yes I am aware that as an “embedded” solution, it may not be viable for some web applications. But from my benchmarking, is faster than MySQL, at least the newer fangled MySQL that includes SQL features that I desire.

Oracle is a monsterous, proprietary morass of unnecessary bloat and complication. Requiring dedicated DBA support and frequent vendor servicing is just not kosher in the 21st century.

If I had to use MySQL, would opt for Postgres instead as the model is much more in the spirit of F/OSS, and again, it’s flavor of SQL includes Oracle power that MySQL lacks.

Finally, there is one you neglected that you might want to explore – CouchDB. Relational model is passe now. Better just to throw objects/object graphs into a data store and retrieve by index. It’s nice to see another option besides relational DB models – early in my career, back when the dinosaurs roamed, our databases on mainframe systems (even going back to the “Burroughs” age) were hierarchical (i.e., IBM IMS) with child/parent or master/slave relationships between data segments. Sort of like XML except all of your schema was defined up front and to change it meant you needed to run an assembler macro as well as write code to “pepper” your revised DB records.

My favorite has to be Access! C’mon, you can even build complete enterprise solutions with it. lol

Ok, seriously now, I kinda like SQL Server. It is simple to use and has enough capabilities for robust solutions. But my vote goes for Postgre now that MySQL is almost a commercial endeavor.

I almost voted for MS SQL Server since that’s what I’ve learned on and I know the most…plus I figured it wouldn’t get much love around here. However, I had to go with MySQL in the end since that’s the one I’ve used the most outside of school.

I’ve used MySQL, SQLite and Oracle, but I learned the later in college and we had to use Oracle Forms Builder, which made me hate every product with Oracle in it :?