Challenges in Implementing Microservices
summary
Argues that microservices succeed only with concrete choices: use an event bus (Kafka cited at ~250,000 msgs/sec) instead of one monolithic DB, give each service a tiny datastore (Redis examples), and favor asynchronous need/solution pattern for robust AB testing and graceful degradation. Describes systems with 300 microservices, Node and Clojure services, Docker and AWS for fast provisioning, and teams deploying at scale (deploy every 3.5 minutes). Lists seven practical challenges and points to Netflix OSS and Elixir as emerging tooling options.
AI-written summary · about
chapters (28)
- 0:00Introduction
- 4:24Hardware Lead Times
- 8:33Project Delivery Cycles
- 9:27How Fast Can You Go?
- 10:40Why: New Problem Domains: The Cynefin Framework
- 13:55Why: Competition is coming
- 17:58Common Theme of All Challenges
- 18:40Challenge 1
- 19:19How Many Databases?
- 21:43Rapids, Rivers, and Ponds
- 22:25High-Performance Event Bus
- 24:50Incremental Applications
- 28:05Challenge 2
- 29:10Chad Fowler vs. Fred George
- 30:13Need Asynchronous Pattern
- 34:24Service Taxonomy
- 35:57Challenge 3
- 36:24MicroServices Like 00
- 36:54Clojure Loves Shared Data
- 38:313 Companies - 3 Variants
- 40:02A New Hope...
- 40:39Challenge 4
- 41:03Open Source Frameworks
- 42:09Challenge 5
- 43:34Challenge 6
- 44:42Challenge 7
- 46:19Count / Size Ratio
- 47:55More Information...
