Scaling PostgreSQL to power 800 million ChatGPT users
By Bohan Zhang, Member of the Technical Staff
Summary
PostgreSQL and MongoDB each excel in different scenarios, and claims that PostgreSQL "cannot scale" oversimplify the reality of database performance.
PostgreSQL supports both vertical scaling (handling hundreds of thousands of transactions per second on a single node) and horizontal scaling through read replicas, native partitioning, and distributed variants like Citus and Aurora. MongoDB's distributed architecture with automatic sharding makes it superior for certain workloads, particularly those with highly variable document shapes and minimal cross-document joins, but PostgreSQL remains the industry's most popular database among developers and has surpassed MongoDB in overall rankings.
Effective scaling depends on matching database architecture to specific workload requirements, team expertise, and operational discipline rather than declaring one technology inherently more scalable than another.
