Fake It to Make It! - TDD of gRPC Microservices
summary
Argues that fast fake frameworks turn the testing hourglass back into a pyramid by replacing slow full k8s E2E runs with in-IDE functional tests. Uses Go, bufconn in-memory gRPC, embedded Postgres per-test fixtures, mockgen and standard go test coverage. Demo shows suites stand up and tear down in ~10 seconds while full KIND/k8s deploys take 30–40 minutes, supports local coverage reporting and optional flip to real cluster; the fake framework is implemented and materially speeds refactor cycles.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:48Fake services for testing
- 3:14Unit vs end-to-end testing
- 6:07Cost of full deployments
- 8:39Lightweight fakes and test doubles
- 12:14Testing pyramid flaws
- 13:18Demo setup
- 14:47IDE testing workflow
- 18:25CLI testing on Kubernetes
- 21:35Coverage for in-memory tests
- 22:40Test data management
- 24:06Testing across stages
- 26:58Summary
- 27:19Q&A
related talks
48mClimbing the Testing Pyramid: From Real Service to Interface Mocks in Go
05mLightning Talk: Unit Testing in Go With K8s Clients & APIs: An Intern’s POV on Why I...
34mAdvanced Testing Techniques in Go
05mLightning Talk: Fake it Until You Make it: Unit Tests with Go-Client Fake Client
20mWriting Microservice Integration Tests in Go (Finally)
25mFakes over Mocks: Shift your paradigm and identify monolithic code
30mFaking It - The Art of Testing Using Verified Fakes
28mIntro to Test-Driven Development in Go