I am wondering if there is a piece of DB theory which clarifies (at the maximum possible) the pros and cons of storing binary data into a Table as a BLOB.

To put it simple, why should I put my images as BLOB and not store them in a directory and have a field with TEXT or VARCHAR where to store the link to every file.?

Consider:
- do you want recovery (of the blob) and integrated recovery(changes in the db vs changes to the blobs)
- if the blob is in the db, then program can get at the db and the blob at the same time, in the same way (SQL)

I make the distinction that the blob is stored in a column, not a table. Eg. Employee Fred has one row and his photo is in the Photo column, like the Address column.

This is an area where the dbs are quite different, so the db you buy will or will not treat blobs well, easily, fast, etc.

Cheers

Here's an online article discussing this point.