Hey guys. I'm working on a project where I'm rewriting an existing website that has 7M monthly views (10m p/ day). I'm used to work with hasura, but this project needs a very custom backend, and each user visit needs to go to the database. So I need to think a bit about performance. When you query hasura through GraphQL, it normally compiles the entire query into a single SQL query to the database, and that's how it gets good performance. There is a lot of info on how prisma is slow on the internet, but I know a lot of them are from prisma 1. How different is prisma 2 performance wise? How does it compare with Hasura now? Knowing that I'll have to write the entire graphql server myself, it means that simply writing resolver for each entity will end up in a lot of queries to the database for just a single graphql query with nested resource, right?

Prisma 2 is just a database client and ORM. It runs the queries you tell it to run. So it is really hard to compare it to Hasura, which takes your GraphQL queries and turns them into SQL queries - as that part is not what Prisma is solving for you.