Change Data Capture (CDC)

Change Data Capture is a technique for detecting row-level changes in a source database — inserts, updates, deletes — and streaming them to other systems in near real time. Instead of re-querying a whole table on a schedule, CDC usually reads the database's transaction log (for example Postgres WAL or MySQL binlog), so it captures every change with minimal load on the source. For SaaS builders, CDC is what keeps your analytics warehouse, search index, or cache in sync with your production database without heavy batch jobs. Tools like Debezium, Fivetran, or Airbyte lean on it heavily. It's the backbone of event-driven architectures and reverse-ETL flows. Practical note: log-based CDC is far gentler than query-based polling, but it needs elevated database permissions and careful handling of schema changes. Also plan for ordering and idempotency — the same change event can arrive more than once, so downstream consumers must tolerate duplicates. Watch replication slot growth on Postgres; an offline consumer can silently bloat your disk.