`findUnique` batching performance

Bug description

I want to retrieve 10,000 accounts (by IDs) from the database.

Due to the nature of how GraphQL resolvers work, I can’t simply use findMany something like

prisma.account.findMany({
  where: { id: { in: [1, 2, 3 /* ... */ } }
})

Instead, I use the findUnique inside of the resolver which gets called 10,000 times:

// inside a GraphQL resolver
prisma.account.findUnique({
  where: { id: account.id }
})

As per https://www.prisma.io/docs/guides/prisma-guides/query-optimization-performance#solving-n1-in-graphql-with-findunique-and-prismas-dataloader, it shouldn’t really introduce a performance bottleneck as the findUnique calls which happen in the same tick should be batched.

From my understanding, it should be almost equivalent to calling the findMany method directly.

Interestingly, if I use the dataloader library in order to create my own batch function (in which I call the findMany method), the performance is almost the same as calling the findMany method directly:

// create a data loader
const accountByIdLoader = new DataLoader<number, Account>((ids) =>
  prisma.account.findMany({
    where: {
      id: { in: ids as number[] },
    },
  })
);

// inside a resolver
accountByIdLoader.load(account.id)

How to reproduce

Full reproduction available on https://github.com/tomdohnal/prisma-dataloader-perf

Running on my machine gives this performance metrics:

Expected behavior

I’d expect the performance of using dataloader library and calling the findUnique method to be similar

Prisma information

Available on https://github.com/tomdohnal/prisma-dataloader-perf

Environment & setup

• OS: Mac OS
• Database: sqlite, postgres
• Node.js version: 12.4
• Prisma version: 2.21.0

[sjerratsch]

Any updates on this? Have there been any releases addressing this issue?

[carloslibardo]

I’m having the same issue, any updates?