Building a hosting platform with Python
summary
A Python PaaS is built from 17 running daemons that communicate with eventlet greenlets for millions of lightweight concurrency (tests at 10M, production ~10k), use ZeroMQ request–reply for resilient message routing and heartbeats to two nexus nodes, and store long‑term structured data in Postgres after having moved Redis to Postgres for complex queries; a custom load balancer replaces a 4,000‑line HAProxy config and improves reloads, subprocesses use a PTY wrapper for interactive commands, and monitoring/logging are captured centrally.
AI-written summary · about
chapters (14)
- 0:00Start
- 3:20Redundancy, Redundancy.
- 4:27Security
- 6:19Basic Technologies
- 7:13Storage
- 9:00Brief Example
- 10:13Standard Classes
- 14:19The Beginning
- 23:13The resource module
- 23:58The signal module
- 25:04The shlex module
- 26:57Don't reinvent the wheel, or optimize prematurely.
- 27:51Reinvent the wheel occasionally
- 29:23Python is really very capable
related talks
39mArchitecture of a cloud hosting service using python technologies: django, ansible and celery
33mLessons learned building Python microservices
25mA microservices architecture with Python and Docker containers
01h 02mUsing Python to Power Serverless Applications
40mPython microservices on PaaS done right
19mPython for Cloud Services and Infrastructure Management
50mBuilding an Application Platform: Lessons from
30mBuilding modern Python web apps with PostgreSQL