High availability
One primary and two replicas across failure domains by default. The CloudNativePG operator watches every instance and promotes a replica automatically when the primary fails.
Read the architectureLoading content
Waiting for the control plane
Platform / Postgres
CheapDB Postgres gives you a fully managed, highly available PostgreSQL cluster on disclosed hardware. Failovers are automated, backups leave the cluster, and billing follows provisioned resources.
Why CheapDB Postgres
One primary and two replicas across failure domains by default. The CloudNativePG operator watches every instance and promotes a replica automatically when the primary fails.
Read the architectureWAL streams off-cluster continuously. Scheduled drills restore selected recovery points in isolated environments and record the result.
Create isolated environments from production backups. Development branches run as cost-optimized single instances and tear down cleanly.
PgDog fronts every cluster with pooling and resilient routing, absorbing application connection spikes while protecting Postgres memory.
Slow queries, execution plans, and schema recommendations are captured automatically. Bind values are redacted inside the database.
Mass-mutation policies block accidental UPDATE or DELETE statements when a WHERE clause is missing.
How it works
When you create an HA cluster, the operator places one primary and two replicas on separate machines. If the primary becomes unavailable, a replica is promoted automatically while the connection layer buffers in-flight traffic.
┌─────────────────────────┐
│ cheapgate pool │
└────────────┬────────────┘
│
╔═══════════╗
║ Primary ║──▶ WAL → R2
╚═════╤═════╝
┌──────┴──────┐
▼ ▼
┌─────────┐ ┌─────────┐
│ Replica │ │ Replica │
└─────────┘ └─────────┘
domain B domain CExtensions
A vetted set of extensions covers scheduling, partitioning, statistics, maintenance, and time-series workloads with tested upgrade paths.
“Our SLA will follow measured failover and recovery results across independent failure domains. The console exposes topology, drill history, and recovery timing throughout that process.”
Single-node development clusters start small, and the console can promote them to a high-availability topology through an online workflow.