ConnectionLess

Challenge of Serverless: Database Connections

For example, with PlanetScale/Supabase/Fauna, you can consider them "connectionless" - connecting pooling isn't an issue, as database queries are made over HTTP! They provide a database as an API. -- A planetscale community discussion

The data revolution along with serverless app revolution, is ConnectionLess

• aka. Data API

Providers

• PlanetScale

  • The main advantage of our database-js driver is that it can be used in serverless environments that don't allow traditional database connections.

• Supabase
• NeonDB
• Deprecated: MongoDB Atlas Data API
• Azure Cosmos
  • https://learn.microsoft.com/en-us/rest/api/cosmos-db/cosmosdb-resource-uri-syntax-for-rest
  • https://learn.microsoft.com/en-us/javascript/api/overview/azure/cosmos-readme?view=azure-node-latest
• upstash

Add-ons

• A list of software that turns your database into a REST/GraphQL API
• serverless-mysql
  • It still requires you to call await mysql.end() to release connection
  • client-side solution: unexpected termination of client still leave server-side zombie connection.
• RESTHeart
  • a Java backend server that provides REST, GraphQL, and WebSocket APIs
  • only support MongoDB as data storage.

Common solution practice

2 tiers connection limit

• When a database reaches the soft-limit we start to terminate the idle connections. We continue to accept new connection requests until the hard-limit is reached. If the database reaches the hard limit then we start rejecting new connections.
• Similar to AWS Lambda approach