MySQL CDC 同步 BigQuery：Binlog 机制有效弥补周期性同步缺陷

近日，托管平台 Erathos 发布了关于 MySQL 变更数据捕获（CDC）同步至 BigQuery 的设置指南，解析了数据同步的原理与配置方法 1。指南指出，传统的周期性批量同步方式会遗漏数据的删除操作和中间更新状态，而采用 binlog（二进制日志）机制能够完整捕获所有数据变更 1。

该指南详细列出了 MySQL 端所需的 binlog 配置、权限要求及注意事项 1。在日志格式方面，二进制日志必须采用 ROW 格式并包含完整的行镜像（FULL row images），且 binlog_row_value_options 参数不得设置为 PARTIAL_JSON 1。在权限分配上，复制用户需要具备 REPLICATION SLAVE 和 REPLICATION CLIENT 权限，以及 SELECT、RELOAD 和 SHOW DATABASES 权限，并且每个复制客户端都必须配置唯一的 server-id 1。此外，binlog 的保留时间必须足够长以覆盖停机时间，默认保留时间为 30 天 1。文章同时提及，Erathos 等托管平台可简化 CDC 管道的搭建与维护，详细的 CDC 设置文档已发布在 docs.erathos.com/connectors/databases/mysql#cdc-setup 1。

Erathos has released a setup guide detailing how to synchronize MySQL Change Data Capture (CDC) to BigQuery using the binary log (binlog) 1. The documentation highlights that traditional periodic batch syncs often miss data deletions and intermediate updates, whereas binlog-based CDC can comprehensively capture all data changes 1.

To enable this synchronization, MySQL must be configured with binary logging in ROW format and FULL row images 1. Additionally, the binlog_row_value_options parameter must not be set to PARTIAL_JSON 1. Regarding permissions, the replication user requires REPLICATION SLAVE and REPLICATION CLIENT privileges, along with SELECT, RELOAD, and SHOW DATABASES permissions 1. The guide also specifies that every replication client needs a unique server-id, and the binlog retention period must be long enough to cover potential downtime, which defaults to 30 days 1. Furthermore, the documentation notes that managed platforms like Erathos can simplify the setup and maintenance of CDC pipelines, with full instructions available at docs.erathos.com/connectors/databases/mysql#cdc-setup 1.