Change Data Capture (CDC) from Oracle to Amazon S3 with AWS DMS

Quick Answer

The two requirements in this scenario, minimizing data loss and supporting change data capture, both point toward the same AWS service because AWS Database Migration Service with ongoing replication is purpose-built to satisfy them simultaneously. Once DMS performs its initial full load of the Oracle source, ongoing replication switches to reading the database's transaction logs and streaming each committed change to the S3 target continuously, rather than relying on periodic batch snapshots that could miss changes made between runs. That log-based approach is what minimizes data loss: because DMS captures every committed transaction as it happens instead of waiting for a scheduled extraction window, there's no gap where new or updated rows go uncaptured between pipeline runs. It also scales comfortably to the 500 GB daily volume described here, since DMS supports parallel replication tasks and task tuning to keep pace with high-throughput sources. The key distinction to hold onto is between a one-time or scheduled batch extraction and true CDC: batch approaches only see the database as it looked at extraction time, while CDC continuously tracks every change as it occurs. Whenever a scenario specifically asks for change data capture from a relational database into S3, along with a requirement to avoid missing changes, that combination is a strong signal that DMS with ongoing, not one-time, replication is the intended answer.