At the moment we are storing all asset information in S3. But S3 has a lot of limitations, is not hierarchical, search is slow, we can't search fast by metadata,.... The key is the identifier, and renaming assets and folders are changing that keys, breaking all links.

Now we need some kind of identifier for each asset, to create/use proper links to them. Later we will need to add more information for each asset, as tags, or any other feature.

S3 should be used only to store the files, not to store asset information needed by magnolia. So, we need to use a database / datasource to store asset information.

The easiest solution is use a JCR application, and just save the files in S3, instead of storing them in Jcr. But like we are trying to remove JCR soon, maybe we can try to find a better database/datasource option.