DBOS conducted comprehensive benchmarks testing Postgres scalability for workflow execution. Their findings reveal a single server supporting 144K writes per second, translating to 12 billion writes daily. This performance exceeds expectations, demonstrating Postgres can handle substantial write-intensive workloads common in durable workflow systems. The tests used AWS RDS db.m7i.24xlarge instances with 96 vCPUs.

Testing different workflow patterns, DBOS found direct workflow processing reached 43K workflows per second. Queued workflows initially performed at 12.1K per second but improved to 30.6K with multiple queues. The bottleneck proved to be Write-Ahead Log flushing, not CPU or IOPS limitations, indicating efficient resource utilization across the database's write operations.

The benchmarks confirm Postgres handles workflow execution at impressive scales. With proper queue distribution, a single server processes billions of workflows annually. For extreme workloads, sharding across multiple Postgres servers offers virtually limitless scaling potential. DBOS has made all benchmark code open-source for developers building reliable systems.