Managed Postgres Databases Across AWS, Azure, GCP — Weekly Benchmarks for May 21 – May 27, 2026

🗄️ Managed Postgres Databases — Weekly Provisioning Benchmarks

Window: May 21 – May 27, 2026 | Total runs: 189

GCP us-east1 had a rough week for managed postgres databases — reliability dropped 9.5 points to 90.5%.

At a glance

Cloud	p50	p95	Reliability	Runs
AWS	4m 58s	5m 29s	100.0%	63
Azure	5m 33s	8m 39s	98.4%	63
GCP	10m 37s	11m 50s	96.8%	63

What changed this week

• ▼ GCP us-east1: reliability 90.5% (was 100.0%, Δ -9.5 pp)

Regional detail

AWS

Region	p50	p95	Reliability
eu-north-1	5m 02s	5m 24s	100.0%
us-east-1	4m 58s	5m 28s	100.0%
us-west-2	4m 58s	5m 29s	100.0%

Azure

Region	p50	p95	Reliability
eastus2	5m 34s	8m 39s	100.0%
northeurope	5m 35s	5m 40s	100.0%
westus2	4m 34s	12m 06s	95.2%

GCP

Region	p50	p95	Reliability
europe-north1	10m 29s	10m 54s	100.0%
us-east1	10m 39s	12m 19s	90.5%
us-west2	10m 37s	11m 50s	100.0%

Methodology

Managed Postgres instance provisioning, measured from create-DB API call to the database accepting a connection. Includes teardown timing for completeness.

All data is collected via ProvisioningIQ — a continuous benchmarking platform that runs synthetic provisioning tests and publishes the results. No estimates, no vendor data; every number above came from a real provisioning attempt.