File access speed vs database access speed

If you're doing read-heavy access (looking up filenames, etc) you might benefit from memcached. You could store the "hottest" (most recently created, recently used, depending on your app) data in memory, then only query the DB (and possibly files) when the cache misses. Memory access is far, far faster than database or files.

If you need write-heavy access, a database is the way to go. If you're using MySQL, use InnoDB tables, or another engine that supports row-level locking. That will avoid people blocking while someone else writes (or worse, writing anyway).

But ultimately, it depends on the data.

It depends on how the data is structured, how much there is and how often it changes.

If you've got relatively small amounts, of relatively static data with relatively simple relationships - then flat files are the right tool for the job.

Relational databases come into their own when the connections between the data are more complex. For basic 'look up tables' they can be a bit overkill.

But, if the data is constantly changing, then it can be easier to just use a database rather than handle the configuration management by hand - and for large amounts of data, with flat files you've got the additional problem of how do you find the one bit that you need, efficiently.

This really depends on many factors. If you have a fast database with much data cached in the RAM or a fast RAID system, chances are bad, that you will gain much from simple file system caching on the web server. Also think about scalibility. Under high workload a simple caching mechanism might easily become a bottle neck while a database is well designed to handle high work loads.
If there are not so much requests and you (or the operating system) is able to keep the cache in RAM, you might be able to gain some performance. But now the question arises, if it is realy neccessary to perform caching under low work load.