About This Architecture

Pigsty is a full-stack PostgreSQL infrastructure platform combining Infrastructure-as-Code (Ansible/Terraform), a Patroni-managed HA cluster with HAProxy load balancing, and a complete observability stack built on Prometheus, VictoriaMetrics, Grafana, and Loki. The architecture routes DBA and developer access through Nginx to Grafana dashboards and pgAdmin, while PostgreSQL replicas stream to pgBackRest for backup and pg_exporter for metrics collection. This production-ready design eliminates manual database administration by automating provisioning, failover, monitoring, and alerting across the entire stack. Fork this diagram on Diagrams.so to customize node counts, add disaster recovery regions, or integrate with your existing Prometheus infrastructure. Pigsty's etcd-based distributed consensus and MinIO object storage ensure data durability and cluster resilience at scale.

People also ask

How do I design a production-grade PostgreSQL high-availability cluster with automated failover, comprehensive monitoring, and backup management?

Pigsty's architecture uses Patroni for HA consensus via etcd, HAProxy with VIP for transparent failover, pgBouncer for connection pooling, and pgBackRest for backup to MinIO object storage. Prometheus scrapes pg_exporter and node_exporter metrics, VictoriaMetrics stores time-series data, and Grafana provides real-time dashboards—all provisioned via Ansible/Terraform IaC.