Hacker News
MySQL CDC to BigQuery: what periodic syncs miss, and how binlog avoids it
Batch syncs miss deletions and intermediate updates and add load to production, whereas binlog-based CDC records every INSERT, UPDATE, and DELETE with full row images. CDC works only with ROW-format binlogs (binlog_row_image=FULL), a replication user granted REPLICATION SLAVE/CLIENT, unique server-ids, and sufficient retention to survive outages. Configured correctly, each change event streams to BigQuery as a row operation, delivering complete, consistent data without snapshot inconsistencies.