Is there a best practice in splitting database files between images and text data?
My thought was that 1 database would have tables with records and other text data. The other database would be records with an index of MD5 hashes and the associated images.
So if a catalog has lots of images, front/back/isometric views of items, there would be no need to update the image catalog if the prices & descriptions of the items are the only thing that change.
Does this make sense? We noticed that images rarely get deleted from our dataset, but, everything else ("meta") is frequently updated.
The goal is to reduce the number of updates for stand alone ("Kiosk") type apps.