prisma-redis-middleware VS aurora

Polyscale supports Cockroach, Postgres, Mysql and MS SQL (AFAIK). You basically create a cache by adding the connection details to the database, and then in Prisma you replace the previous connection string with the one supplied by Polyscale. I just set it up very recently though, but the speed differences are incredible. I think Planetscale Boost and Prisma Accelerate are similar alternatives. There is also a recent open-source Prisma middleware which uses Redis to cache all your Prisma queries, but I didn't try it yet: https://github.com/Asjas/prisma-redis-middleware

There's also a middleware library that easily lets you cache in Redis: prisma-redis-middleware

Yes, there are a few ways to achieve this! It's not supported natively yet but in this issue lots of people list their favorite workarounds which you can draw inspiration from. One of the most popular ones is to use a library written by one of our community members (who we actually now hired as a Developer Advocate :D) called aurora.

Splitting the schema file is indeed not yet possible but we will be working on this soon. Also, there are already a number of tools from the community that allow this (e.g. Aurora) and some people just use simple bash scripts to concat the .prisma files before running CLI commands against them (the GitHub issue lists a lot of workarounds).