Does Postgres Scale?

Benchmarks on an AWS RDS Postgres instance with 96 vCPUs, 384 GB of RAM, and 120K provisioned IOPS found that a single server can sustain up to 144,000 small transactional writes per second, equivalent to roughly 12 billion daily writes. Tests of durable workflows, which require database writes to record starting and completion states, reached 43,000 workflows per second for direct execution, or about 86,000 writes per second, with larger tables and indexes reducing throughput relative to raw inserts. In both write-heavy tests, write-ahead log flushing was the main limitation because Postgres serializes committed writes through a single WAL. Postgres-backed queued workflows initially peaked at 12,100 workflows per second due to lock contention around queue-head rows, but distributing work among multiple queues or partitions increased performance to 30,600 workflows per second, after which WAL throughput again became the constraint. The results indicate that a single Postgres server can support substantial durable workflow volumes, while sharding across servers can provide additional capacity for larger workloads.

No tracked trend matches for this post yet.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.