One of the reasons I'm asking for help testing is there are so many variables with image files and computer configurations, and so many individual needs.
One of my tests was pointing Photo Sieve at a directory with 16,000 sub-directories and 41,000 images. The way I use it is a multi-step process:
1) Photo Sieve builds a directory tree of all sub-directories containing images. This can probably process a very large drive since this step does not load images or create thumbnails.
2) Apply one or more sieving operations based on location, take taken, file name or file date.
3) View and select images from each individual sub-directory. The number of sub-directories is likely greatly reduced by previous sieving operations.
4) Load selected images. By this step, the 41,000 images should be narrowed to a few hundred related images.
5) Create custom tags or text. If all the loaded images are related in some way, this may be a single paste operation.
6) Optionally save the loaded images to an album. Multiple albums may contain the same image. Since tags and text are stored in the jpg files, edits to a picture automatically appear in all albums.
7) Repeat for a different group of related images.
8) Later sieving operations may include added custom text or tags.
There are also other ways to divide and conquer when working with many images.
With so many backup tools available, I'm not inclined to create another. Are you familiar with Beyond Compare? It displays 2 drives/directories side-by-side and has multiple select, compare, and copy options. It can also display difference between 2 pictures. It will likely work well for the backup scenario you described.