Strangling the Monolith With a Data-Driven Approach
summary
A health‑insurance cost estimator is rewritten by routing live traffic through a proxy called Project X that logs every request and response, compares legacy and new calculator outputs, and flags mismatches; engineers run a baseline variance test in CI (Concourse) using real production pairs, track 90% matched cases, apply a financial‑risk formula, and progressively cut over. The team delivers reusable microservices, slashes QA effort, and completes the decisive strangling of the monolith in about 13 weeks.
AI-written summary · about
chapters (19)
- 0:00Introduction and speakers
- 1:05Case study overview
- 2:10Legacy system challenges
- 4:38Initial rewrite struggles
- 8:31Pivoting to data-driven approach
- 9:54Strangler pattern explained
- 13:10Project X and data logging
- 14:55Comparing old and new systems
- 16:39Automating error detection
- 17:20Tracking business metrics
- 19:27Handling monolith inaccuracies
- 20:30Strangling false positives
- 21:58Monolith shutdown decision
- 23:26Losing the baseline confidence
- 24:56Self-baselining production
- 26:01Baseline variance testing
- 27:50Key learnings summary
- 29:18Pattern comparison recap
- 32:09Q&A
related talks
32mQuietly Strangling the Monolith - a real-world migration of a high traffic monolith
20mSplitting the Monolith
44mDemonolith the Monolith Think Again!
39mWays out of a monolith
26mAccept Complexity or Reject It? — Reflections Facing a Monolith Raised Alongside Business Growth
28m193 Easy Steps to DevOpsing Your Monolith
40mWhy your best engineers should look after the worst systems
33mFireside chat - Monoliths, Microservices & Modernization – Practical Strategies