CDC replication backbone
Log-based change data capture streams every row change out of the OLTP database — feeding search, caches and the warehouse without touching app code.
What this design does
The database's write-ahead log is already a perfect changelog — CDC just reads it. No dual writes, no polling, no app changes: every committed row change becomes an event, validated against a schema registry, fanned out to whatever needs fresh data. The outbox table rides the same pipe for app-level intent events.
The patterns it composes
- Log-based CDC — reads the WAL — captures every commit, misses nothing, adds no app code
- Transactional outbox — intent events commit atomically with business rows, shipped by the same CDC
- Schema registry & evolution — producers can only publish compatible schema changes
- Streaming ETL — transforms run on the stream, not in nightly jobs
- Dead-letter channel — unparseable changes park instead of stalling the backbone
Import it, attach your own numbers, and verify the design against failures — or read the engineering notes.