How Not to Screw Up Moving Processing from 1 Country to 33
summary
Describes scaling a payment/processing system from serving one country to 33 by splitting monoliths into independent microservices for queue management and intelligent (ML) processing, unifying per-country processing algorithms, removing MySQL-based queues, automating deployments, adding service discovery and load testing, and achieving reliable handling of hundreds of thousands of events per day.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 10:17Monolith scaling problems
- 13:55Splitting to microservices
- 23:58Service dev standardization
- 25:47Performance and load testing
- 31:48Model testing and integrations
- 33:16Traefik discovery
- 36:12Refactoring results
- 38:22Conclusion
related talks
43mMicroMonolith - Top anti-patterns of distributed systems
42mQ&A - Monolith to Microservices
32mSplitting the monolith
37mThe Need For Speed: Scaling Go Microservices - Artur Kondas
49mPayment System: Survival Guide
34mMigrating a High‑Load Project to Microservices
31mMonolith to Microservices: A Guide to Seamless Transitions
40mUncoupling Systems
