Using the right tool

I spent about 12 hours crawling over Google searches, Usenet archives, and a
great deal of GIS matter without finding much in the way of solid answers.
Perhaps someone on this list can provide a bit of insight.

I'm in the need of cataloging a growing collection of USGS satellite
imagery. That in itself has been handled rather handily using flat files,
but with a recent decision to acquire non-USGS data (potentially tens of
thousands of images) covering geographical regions, the need for a reliable
catalog is paramount. You can imaging how easily this would become a
management nightmare. This data will be tagged, cataloged, and filed by me,
rather than the easy-to-handle USGS data that has a common packaging and
naming style.

Most people who create an image library are doing so for web-based
applications, usually, it seems, using MySQL. While I have no convictions to
either database (it's just a tool, so the end is more important than the
means), I've decided to use Postgres since it has strong ties to the GIS
community.

But for this task, I've encountered little in the way of stories from users
who have used Postgres to catalog image data, despite fairly intense
digging. The thing is, these are not small images by any means. So what
works well for a few thousand < 250k JPEGS might not work so well for a few
thousand 25-125 meg TIFFs and SID files.

Can anyone suggest some references for storage of binary objects of this
magnitude (100 gig)? While it isn't particularly important to store the file
in the database itself, it sure would simplify things -- that way if the
file is moved, the db link isn't broken. Rather than reinvent the wheel and
build a database and interface solution, is any information available on
implementations such as this?

Regards,
Nathan Hopper