Deploying on AWS Lambda

It is possible to deploy Platformatic applications to AWS Lambda by leveraging @fastify/aws-lambda.

Once you set up your Platformatic DB application, such as following our tutorial, you can create a server.mjs file as follows:

import awsLambdaFastify from '@fastify/aws-lambda'
import { buildServer } from '@platformatic/db'

const app = await buildServer('./platformatic.db.json')
// You can use the same approach with both Platformatic DB and
// and service
// const app = await buildServer('./platformatic.service.json')

// The following also work for Platformatic Service applications
// import { buildServer } from '@platformatic/service'
export const handler = awsLambdaFastify(app)

// Loads the Application, must be after the call to `awsLambdaFastify`
await app.ready()

This would be the entry point for your AWS Lambda function.

Avoiding cold start

Caching the DB schema

If you use Platformatic DB, you want to turn on the schemalock configuration to cache the schema information on disk.

Set the db.schemalock configuration to true, start the application, and a schema.lock file should appear. Make sure to commit that file and deploy your lambda.

Provisioned concurrency

Since AWS Lambda now enables the use of ECMAScript (ES) modules in Node.js 14 runtimes, you could lower the cold start latency when used with Provisioned Concurrency thanks to the top-level await functionality. (Excerpt taken from @fastify/aws-lambda)