Decomposing a live application with confidence
summary
Company decomposes a live, profitable monolith by piping duplicated feedback into Kafka, applying event storming and context mapping, and building streaming jobs that emit a unified schema into new topics; tooling includes Kafka streams, Kubernetes for ops, Chimera PHP (tactician + zend-expressive + Symfony DI), phpunit/phpstan/infection. A refactor raises coverage from 3% to 25%, exposes past fragility (one refactor once sent 50,000 emails), and proves gradual strangling, metrics/SLOs and consumer groups let new services roll out safely.
AI-written summary · about
chapters (26)
- 0:00Intro
- 0:51Bassano del Grappa 1
- 2:03Live software!
- 3:27No tests
- 3:40Outdated stack
- 4:22Random bugs
- 5:28Saved searches
- 7:37Quality increase
- 7:55Remember?
- 8:20Sometime ago...
- 9:09Big ball of mud
- 10:10Strangler application
- 14:12Decomposing
- 15:53Domain knowledge
- 16:06Event storming
- 16:38context mapping
- 17:22Gherkin stories
- 18:28Microservices
- 19:34Evolvability
- 23:06Strictness
- 24:17Growth & scale
- 32:53Feedback collection
- 35:29Lessons
- 35:57Domain exploration
- 36:27Communication
- 36:45Discipline
related talks
57mSplitting Monolith Application to Microservices
45mUsing modern .NET technologies to decompose a monolith into micro-services
32mSplitting the monolith
40mFrom Monolith to Observable Microservices using DDD
41mReducing Microservice Complexity with Kafka and Reactive Streams
43mBreak Up the Monolith: Testing Microservices
52mTools to Slay the Fire Breathing Monoliths in Your Enterprise - Kelapure, Anderson, Szodfridt
21mHow Apache Kafka® Generates Value In Retail With Confluent