Well I'm currently using the file system for large files. However because
of that I can see a few reasons why people might want to use Postgresql to
handle them. Others can probably mention more.
Using Pg to handle large stuff makes more consistent overall and it's
easier for you to handle exceptions - e.g. if things fail the whole thing
is rolled back, and you theoretically don't get the large files dangling
around. Well ok you probably do until the next vacuum, but at least you
don't have to write your own vacuum to handle that ;). Basically you shift
the problem to Pg (and the very fine developers :) ).
The reasons I decided to go file system were:
1) I'm using Linux and ext2fs has a 2GB limit on files, and it seems like
6.5.3 tables are stored as single files, so better not go down that path :).
2) I'm using Perl, DBI etc and a brief look at BLOB handling put me off.
Maybe it was unwarranted, but given 1) I decided to call the whole thing off.
Cheerio,
Link.
