PgDog 融资：Postgres 水平扩展代理
值得看指数 63.0 NO. 011 · 2026.06.11
发布2026/06/10Score294Comments155
为什么值得看
PgDog 是一个 Postgres 代理层，通过透明分片让单库水平扩展到 100TB+ 和百万 QPS，已处理生产环境 200 万 QPS。对不想迁移到专用分布式数据库的团队，这可能是保留 Postgres 生态的最小改动方案。
编辑判断
Postgres 分片不是新需求，但现有方案要么侵入性强（Citus 改查询语法、YugabyteDB 换存储引擎），要么运维重（手动分片+应用层路由）。PgDog 走透明代理路线，对应用零改造，这直接对标了几年前 PlanetScale 对 MySQL 做的事。
关键风险在于：透明分片的查询路由和跨分片事务一致性很难做好，尤其是复杂 JOIN 和聚合。PgDog 声称已跑 200 万 QPS，但 HN 讨论里有人质疑其分片策略细节未公开。如果查询模式复杂，建议先拿只读分析场景试水，而非直接替换核心交易库。
社区反馈
意见分歧 140 条评论
核心争论：PgDog 能否在保持 Postgres 生态的同时，真正替代专用分布式数据库的扩展能力
>PgDog is a sharder, connection pooler and load balancer for PostgreSQL. Written in Rust, PgDog is fast, reliable and scales databases horizontally without requiring changes to application code. Still trying to figure out how this works technically, is the performance gain really just re-write in ru
Not quite. The performance gain is to bring those features to Postgres! Edit: Performance gains are from having the ability to load balance reads (horizontal scaling for read queries) and scale out writes (with sharding). Once instance bottleneck in Postgres has many faces: 1. Behind schedule vacuum
Oh thanks for clearing that up.