Legacy Architecture Migration Patterns
summary
Describes concrete migration choices: PayFit keeps legacy and new systems running for months, discovers 21 entry points for creating employees, and chooses a read-first migration with a risky threedirectional sync to keep three databases consistent (Mongo, Postgres, event store). Builds an observability portal, pushes events into Snowflake, uses DLQs and an AWS integration platform, and republishes legacy events into a new model; republishing unlocks dependent teams but requires careful mapping and operational work. Half of manual creations move to the new system after a year.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:25Legacy migrations matter
- 5:44Modernizing principles
- 8:54Strangler fig patterns
- 15:06Case study employee migration
- 23:26Data sync challenges
- 26:14Consolidating entry points
- 30:53Case study contract migration
- 35:50Legacy event reliability
- 37:40Key takeaways
- 38:45Final thoughts
- 39:07Q&A
related talks
01h 01mLegacy Architecture Migration Patterns with DDD
20mLegacy Migration... Differently
43mPatterns of Legacy Displacement
47mLegacy Software: Really a Problem?
18mMigrating Legacy Financial Systems to Cloud: Challenges and Best Practices
31mWhen Bad Architectures Happen To Good People: Migrating Legacy Systems To Serverless
44mLegacy Software Really a Problem
58mJourney to the Cloud - a Rearchitecting Story