What is change data capture?
Change data capture (CDC) is a method of identifying the rows that have changed in a source database and delivering just those changes, in order, to another system. Most implementations read the database’s transaction log, so they see each insert, update, and delete without adding load to the application or requiring a last-modified column in every table. The changes are published as a stream of events that a data pipeline, a lakehouse table, or a search index can consume within seconds. CDC replaces the older approach of copying whole tables on a schedule: instead of reloading five million customer rows each night to pick up the two thousand that changed, the pipeline applies two thousand events.
Why change data capture matters
CDC is how analytics and AI systems keep up with operational systems without slowing them down. A retailer’s inventory agent that reads from a lakehouse table fed by CDC sees a stock adjustment seconds after the store associate scans it, rather than the next morning. Deletes are captured too, which matters for privacy requests: when a customer record is erased in the CRM, the erasure propagates to every downstream copy.
What to watch for
- Confirm the source database and hosting arrangement allow log access, because some managed and SaaS systems do not expose it.
- Plan for schema changes at the source, which break consumers unless the CDC tool forwards them as events.
- Keep an initial full snapshot process, since CDC only carries changes made after it starts.