Beyond Strangler Fig: Effective Patterns for Architectural Modernization
summary
Describes a migration where a German payment provider carves out a fraud engine into a standalone REST product, discovers writers are embedded in the legacy DB, and avoids a weekend big‑bang by using change data capture to emit events, rebuild read models, and incrementally shift readers and writers. Uses event inception, feature flags, canary releases (10%→20%→100%), dark launches and the Scientist/Legacy‑mimic patterns to verify correctness and roll back safely while measuring fitness functions (e.g. incidents 20→5/week, faster onboarding).
AI-written summary · about
chapters (16)
- 0:00Introduction
- 2:12Why modernization matters
- 5:00Origins of legacy systems
- 8:32Incremental change over big bang
- 11:45Target state and progress
- 15:56Finding boundaries
- 18:44Business capabilities vs journeys
- 23:14Modularization vs distribution
- 25:21Strangler Fig limitations
- 27:04Incremental data migration example
- 32:42Transactional architectures
- 33:45Canary releases and dark launches
- 34:48Scientist pattern
- 36:53Dual transformation
- 39:02Modernization principles
- 41:13Q&A
related talks
47mRefactor Without Breaking Everything: Anatomy of Incremental Modernization Patterns
43mPatterns of Legacy Displacement
46mStrangling the dragon: modernizing legacy systems with the Strangler Fig pattern
58mLegacy modernization made practical: A live coding example of modernization patterns
59mArchitecture Modernization Anti-Patterns: The Cost of Good Intentions
24mModernise your monolith using strangler pattern
15mBol.com and the Accelerated Legacy Challenge
43mLegacy Architecture Migration Patterns