WhoBuilt

Fastest Way to Deploy a Database

Jul 10, 2021· 12 min· 15K views·Watch on YouTube

Lee Robinson (leerob, role and company not stated) compares how quickly 13 database services let users get from their home page to a provisioned database.

What the comparison measures

Robinson times the steps from a service’s home page through login and database deployment to having a usable database. He says the test is informal because providers use different sign-in methods and offer different database types, but treats onboarding speed as one indicator of developer experience.

Trying the providers

The tests cover PlanetScale, Railway, Supabase, Firebase, Nhost, Heroku, DigitalOcean, Render, AWS, Google Cloud, Fauna, Upstash, and MongoDB Atlas. Some databases were ready quickly, while others required provisioning waits; Google Cloud’s setup ran into billing and project confusion, so Robinson marked it as disqualified.

Provisioning and setup friction

Supabase asked Robinson to wait while it built the database, and DigitalOcean and Render showed provisioning in progress before completion. Heroku’s Postgres add-on flow was confusing, while AWS and MongoDB Atlas also required waiting; Robinson noted he needed to delete the AWS database to avoid charges.

PlanetScale wins the timed test

PlanetScale took 10 seconds and won the comparison for time to database, using Vitess, its open-source SQL solution. Robinson cautions that the results compare unlike services and database types, so the experiment is mainly a look at onboarding experience rather than a scientific benchmark.

Takeaways

  • Treat time to a usable database as one practical measure of onboarding quality.
  • Make provisioning status and expected wait times visible during setup.
  • Check billing requirements and remember to remove paid resources after experiments.
  • Compare onboarding flows with the understanding that database types and setup steps differ.

Mentioned in this talk

Search inside this talk

What was claimed 8

Related talks