Microliths: how to avoid traps in your stack and your culture
summary
Claim: moving a monolith into hundreds of services only helps if synchronous coupling is removed; tools used are Kafka, Spark, ELK (Elasticsearch Logstash Kibana), Consul/Nginx, Docker/UCP and a homegrown CLI called Taylor. Metrics: platform runs 90+ services, ~thousands of container instances, 40 million active users and 30,000 logins/sec and cuts a typical business transaction from eight over‑the‑wire calls to one. Built: a standardized event messaging pipeline, transaction tracing and log analytics; result is reduced chatter and faster, observable deployments with ongoing governance needs.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:40What is a Microlith
- 1:50Key takeaways
- 2:40About Leafeon
- 4:40About Livefee
- 6:15Culture
- 11:00Services
- 12:15Phase 2 Governance
- 14:00Phase 2 Architecture
- 15:57Metadata
- 18:30Logging analytics
- 19:45Security
- 21:07Logging
- 23:14World definition
- 23:51Taylor
- 24:40Dev Build Pipeline
- 27:08Conclusion
related talks
19mMultigrain Services
53mHow do we build high‑load microservices
57mSplitting Monolith Application to Microservices
39mGrow with Less Pains - Meshing From Monolith to Microservices
30mMonolith to Microservices
35mLegacy Monolith to Microservices
30mMonolith-ifying perfectly good microservices
48mMicroMonolith