Redpanda Connect now ships a native `oracledb_cdc` input that reads Oracle redo logs via LogMiner to capture row-level changes without Debezium, Kafka Connect, or a JVM. The walkthrough builds a two-stage pipeline: Oracle changes flow into a durable Redpanda topic, then a second pipeline reads that topic and writes to ClickHouse using `sql_insert` with the `ReplacingMergeTree(_scn)` engine to automatically deduplicate updates. Both stages are configured as single-binary YAML files run with `rpk connect run`. Key details include enabling ARCHIVELOG mode and supplemental logging on Oracle, using `stream_snapshot: true` for initial data load, setting Kafka keys to primary keys for ordered partitioning, and batching inserts for ClickHouse performance. The `oracledb_cdc` connector is an enterprise feature available via a free 30-day trial license.
