I think this discussion about database vs file systems is kind of lost. A database is too ill defined to get any real traction.
If we are talking about a classic database, like PostgreSQL, Oracle, DB2, etc, then I will say that these systems "can" store large data objects, but it is not optimal, and if you are interested I can point you to some resources. There are "databases" designed to hold large data objects, but that is a different topic. >From a logical point of view, a file system is a type of database. It maintains data, maintains meta data, can be queried. You can even store data with a file in some file systems with extended attributes. It comes down to what the system is designed to do. A "classic" database doesn't necessarily need a file system (oracle). Its job is to maintain data integrity and relations. They are not optimal for "storage" but are designed to be "right." A file system is designed for storage and retrieval. > Rich Pieri wrote on 2026-01-24 06:20: > >> Stream objects [x-rays & PDFs] can be stored in global nodes but the >> typical use and best practice is to store these data as files on the >> filesystem. > > Sounds good, attachments as files isn't a Bad Idea. > > Of course, neither is email as database records. > > > I'd have some hesitation with bifurcating the storage between a DB and > the file system do to the possibility (inevitability?) of the 2 storage > systems going out of sync. > > It's fine for web apps, but email messages (and medical records!) are a > bit more important. > > > > If a file gets deleted, how does the DB know to update its index? Is all > file access through the one software? That's reasonable to expect, but > also one would want occasional re-indexing probably, should such a > situation arise. > > > For example, NextCloud runs a cron job to look for changes in the users' > directories in the file system and re-index folders with changes. > > >> Why? Because storing them in the database fills the in-memory index >> with unneeded bloat which slows down the process of data retrieval. > Mime-types to exclude indexing the contents of several formats of > attachments should alleviate this, no? > > _______________________________________________ > Discuss mailing list > [email protected] > https://lists.blu.org/mailman/listinfo/discuss > _______________________________________________ Discuss mailing list [email protected] https://lists.blu.org/mailman/listinfo/discuss
