Lessons learned building Python microservices
summary
Dozens of microservices on Kubernetes in Java and Python talk to an InfiniMix legacy DB and serve 800 branch terminals; contracts are enforced with pact contracts and a homegrown pactman Python verifier. Python services run Django + mod_wsgi and DRF, use git submodules for simultaneous library edits, pytest/black/flake8 for consistency, retries plus a DB select 1 validation to cut intermittent errors, profiling cuts a slow 400ms path down to ~50ms, and caching drops 40,000→4,000 calls.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:56Agenda
- 2:08Architecture
- 6:00Git sub modules
- 7:09Consistency of environment and practices
- 9:56Testing
- 11:02Consistency of editor
- 13:01Interoperability
- 16:48Resilience
- 17:49Database Retry
- 21:46Intermix
- 22:42Django
- 23:47Django Rest
- 24:50Identity Columns
- 26:41Performance
- 29:26Reducing SQL queries
related talks
44mPython in the MicroServices world
34mPython Micro-service Architecture
30mHow To Build a Python Microservice Without Losing a Job
25mA microservices architecture with Python and Docker containers
40mPython microservices on PaaS done right
42mBuilding a hosting platform with Python
29mPrototyping Python Microservices in Production
18mMicroservices: the small print