I’m experimenting with running Strapi on relatively small hardware, and the performance is more than adequate for even larger numbers of reads so far
However there are a lot of use cases like larger volume batch jobs that I’d much rather implement in separate jobs/containers, and in more performant languages. In these cases it would be optimal to go to the DB directly rather than feeding potentially 10s of thousands of requests (coming from a guaranteed safe source and in good format) through the REST API.
Looking at the tables managed by Strapi, it didn’t immediately look to me like there would was any ‘magic’ going on that could interfere with this - the DB structure is straightforward and if I create individual records they seem to work fine in the Strapi UI.
Has anyone had any experiences with this? Did you run into any issues? Are there any official views on whether this is encouraged/discouraged?