Just over 1 month ago, Deno has announced support for NPM, which means that 1.3 million new modules can be used to build a Deno application. Also, it means that a lot of people can migrate from Node to Deno, just by changing some configurations and imports.

This got me thinking about moving my application, Drakkle, to Deno, but I couldn't find any benchmarks on GraphQL, so I decided to do some simple tests.

Instead of choosing the most popular stack (Apollo Server + Express), I decided to choose the most performant: Fastify + Mercurius.

Coding

I love Typescript, but since it's a simple benchmark, Javascript seemed like a good enough choice.

Using the same example from the Mercurius documentation, an add query was created, which returns the sum of two numbers passed.

The code for Node looks like this:

import Fastify from 'fastify'
import mercurius from 'mercurius'

const app = Fastify()

const schema = `
  type Query {
    add(x: Int, y: Int): Int
  }
`

const resolvers = {
  Query: {
    add: async (_, { x, y }) => x + y
  }
}

app.register(mercurius, {
  schema,
  resolvers
})

app.listen({ port: 3000, host: '0.0.0.0' })

The code for Deno was the same, except for the imports:

import Fastify from 'npm:fastify'
import mercurius from 'npm:mercurius'

Benchmarking

Using autocannon, I did the following script to simulate 500 concurrent connections over 30 seconds:

npx autocannon -c 500 -d 30 -m POST --body 'query Add { add (x: 50, y: 50) }' -H 'Content-Type:application/graphql' localhost:3000/graphql

Node

199k requests in 30.23s, 37.9 MB read

Deno

93k requests in 30.36s, 16.9 MB read

Conclusion

As you can see, Node it's just over 2 times faster than the Deno.

To try this further, I tried to compare by creating a fork of the process for each CPU, using Node's cluster module, but this module has not yet been ported to Deno.