itsual Change Data Capture
Overview
CDC turns database changes into a stream of events, enabling downstream systems to react without dual writes or heavy polling.
When to Use
- Feeding search indexes, caches, or analytics from a source of truth database
- Building event streams from existing applications with minimal application changes
- Integrating multiple systems around shared data
Key Considerations
- Source database support and impact (triggers, transaction logs, etc.)
- Ordering and consistency guarantees
- Schema change handling
- Exactly-once vs at-least-once delivery to downstream consumers
- Initial snapshot + ongoing changes
- Monitoring lag and failures
Principles
- Prefer log-based CDC over trigger-based when available and appropriate
- Design consumers to be idempotent
- Make lag and error states highly visible
Verification
- Changes are captured completely and in order (within documented guarantees)
- Downstream systems can catch up after failures
- Schema evolution does not silently break the pipeline