This example uses the Serverless framework to deploy a Dgraph instance to EC2 and DgraphQL endpoint to Lambda and API Gateway.

First, setup Serverless with your AWS identity.

Next, update provider.vpc.subnetIds in serverless.yaml with the subnets you want your Lambda and Dgraph instanc to run in. You can list the subnet IDs in your account from the command line with the AWS CLI:

$ aws ec2 describe-subnets | grep SubnetId

You might also want to change the region the service will run in (it is set to eu-west-2 in serverless.yaml)

Warning: I hit memory limits running Dgraph on the free-tier t2.micro instance, so it's currently configured to run on t2.medium which is not free. It will cost you about $1.20 a day so don't forget to run yarn run remove when you've finished experimenting

Finally, install the dependencies and deploy the service:

$ yarn install
$ yarn run deploy

You can then open the deployed endpoint in your browser to interact with the service through GraphiQL.

Here's a mutation to get you started:

mutation {
  createPerson(input: {
    name: "John Lennon",
    friends: [
      { name: "Paul McCartney" },
      { name: "George Harrison" },
      { name: "Ringo Starr" }
    ]
  }) {
    person {
      id
      name
      friends {
        id
        name
      }
    }
  }
}

After changing src/schema.graphql run yarn run deploy again to update the deployed schema.

For more on what you can do with the schema checkout the Dgraphql documentation.

To remove the service run:

$ yarn run remove

Serverless seems to have an intermittent issue cleaning up network interfaces. If the stack removal stalls you might have to remove the interface manually in the AWS EC2 dashboard and try removing again.

Happy hacking :)