Death to Test Environments
summary
Argues that dedicated test environments usually waste time and fail to reproduce real data or usage patterns, citing a healthcare case where on‑prem RabbitMQ versus cloud Azure Service Bus created false parity and extra complexity. Recommends testing in production with mark test users to hit real datasets, using Stripe test keys per user, applying contract testing (Pact) to block incompatible deploys, and relying on feature toggles and faster CI/blue‑green deploys so teams become expert at shipping and fixing in production.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 0:30What is a test environment
- 1:33Conways Law
- 1:44Challenge
- 2:09Resources
- 2:41Better rigged
- 4:14Traditional example
- 7:00Horizontal test environment
- 8:19Contract testing
- 8:50Feature toggles
- 9:10Work together
- 9:32Improve delivery time
- 9:50QA environment
- 10:07Smoke detector
related talks
01h 02mThe Art & Joy of Testing in Production
45mEnd-to-End Test Architectures, a Dead End Road
32mI Test In Production
57mAdvanced Testing Techniques: Tips from the trenches
38mReal-Life Solutions for Test Environment Challenges
36mTesting In Production
44mDo You Really Need a Test Environment?
22mTesting in Production