Change Data Capture

CDC Streaming Database

Capture changes from your production database in real-time. RisingWave enables zero-latency pipelines with no dual writes, no log parsing, exactly-once guarantees.

Start Free Learn More

10-50ms

PostgreSQL CDC

Lowest latency CDC ingestion directly from PostgreSQL WAL logs

Exactly-Once

Delivery Guarantee

No duplicates, no data loss — automatic consistency across the pipeline

Zero Code

No App Changes

CDC captures changes from the database log — no dual writes needed

SQL

Transform + Serve

Join, aggregate, and enrich CDC streams using standard SQL

Why CDC

Why is CDC better than the dual write anti-pattern?

Dual writes force your application to write simultaneously to the database and a message queue. If one write succeeds and the other fails, data becomes inconsistent and debugging becomes a nightmare. CDC treats your database as the single source of truth, automatically capturing changes with no application changes required.

Factor	Dual Writes	CDC (RisingWave)
Consistency	Risk of partial failure	Automatic, guaranteed
App Changes	Modify every write path	Zero — reads from DB log
Complexity	Two write targets to manage	Single source of truth
Failure Mode	Silent data divergence	Checkpoint-based recovery

→PostgreSQL CDC with 10-50ms latency via WAL log reading
→MySQL CDC with 50-100ms latency via binlog connectors
→MongoDB CDC with 50-100ms latency via change streams
→Oracle and SQL Server CDC via Debezium connectors

Use Cases

What real-world use cases benefit from CDC streaming?

Real-time replication keeps data lakes in sync with source databases. Streaming analytics lets you analyze customer behavior as it happens. Cache invalidation automatically refreshes caches when data changes. Feature stores for ML models always contain fresh data for training and inference.

Real-Time Replication

Mirror database to data lake. Always in sync. No batch exports or scheduling.

Streaming Analytics

Analyze customer behavior as it happens. Create real-time materialized views on CDC data.

Cache Invalidation

Detect DB changes, invalidate cache automatically. No stale data served to users.

Feature Store

CDC into RisingWave, ML models always see fresh features for training and inference.

How It Works

What is Change Data Capture and how does it work?

Change Data Capture (CDC) captures every INSERT, UPDATE, DELETE operation on your database in real-time, creating a stream of changes. Unlike batch exports that run hourly or daily, CDC provides a live feed of data modifications as they occur in your source database.

→Reads database transaction logs (WAL, binlog) without impacting production
→Streams every INSERT, UPDATE, DELETE as it happens
→No batch exports, no scheduling, no extraction windows
→Automatic schema evolution as source tables change
→Exactly-once delivery guarantees with checkpoint-based recovery
→SQL-based transformations on CDC streams before downstream delivery

Frequently Asked Questions

What is Change Data Capture (CDC)?

Why is CDC better than dual writes?

Which databases does RisingWave CDC support?

Can I use CDC for real-time analytics?

Ready to enable real-time pipelines?

Start capturing database changes in real-time with SQL.

Enable Real-Time Pipelines

PostgreSQL Streaming →Stream Without Java →