Use Cases
Pigsty is built around PostgreSQL, but it's far more than a "database installer". The categories below help you decide quickly: is Pigsty right for me?
🏢 Scenario 1 — Self-hosted open-source RDS
Pain: managed RDS is expensive, versions lag, extensions are restricted, data residency is constrained.
With Pigsty:
- Deploy a production HA cluster on ≥3 nodes in your own IDC / hybrid cloud / VPC
- Automatic failover via Patroni + etcd; HAProxy handles read/write routing
- pgBackRest provides PITR; pair with MinIO for off-site cold backup
- Grafana dashboards and Prometheus alerts out of the box
Further reading: Deployment guide → Architecture planning → Three-node HA template
🤖 Scenario 2 — AI / vector retrieval backend
Pain: LLM apps need vector search, semantic cache and memory storage — adding Milvus / Chroma increases operational burden.
With Pigsty:
- Ships with
pgvector,pgvectorscale,pg_search,pg_trgmetc. — one engine handles vector + full-text + JSON search - The VIBE module offers a Code-Server / JupyterLab / Claude Code sandbox for AI development
- FERRET gives PostgreSQL a MongoDB-compatible wire protocol for LangChain / LlamaIndex integration
Further reading: PGSQL extensions → VIBE module
🧪 Scenario 3 — Local development / data sandbox
Pain: developers want to reproduce a production-like PostgreSQL environment locally with the full monitoring stack and rich extensions.
With Pigsty:
- One command brings up Pigsty on a 1C/2G Linux VM or WSL — PostgreSQL, Grafana, Prometheus included
- Vagrant / Terraform templates provision multi-node sandboxes in minutes
- 40+ configuration templates (meta, rich, fat, slim, infra, vibe) ready to reuse
Further reading: Single-node install → Sandbox → Config templates
📊 Scenario 4 — Multi-model data platform
Pain: the business needs OLTP, time-series, geo, document, KV and object storage simultaneously — you don't want to introduce a separate cluster for each.
With Pigsty: use PostgreSQL as the core; one infrastructure handles:
| Data type | Stack |
|---|---|
| OLTP | PostgreSQL 17 + Patroni HA |
| OLAP | Citus / DuckDB FDW / pg_mooncake |
| Time-series | TimescaleDB |
| Geo | PostGIS |
| Vector | pgvector / pgvectorscale |
| Document | FerretDB (MongoDB protocol) |
| KV / cache | Redis cluster / sentinel / master–replica |
| Object storage | MinIO single / multi-node |
| Distributed FS | JuiceFS |
Further reading: Module overview → Redis → MinIO
🔒 Scenario 5 — Compliance / private / domestic stack
Pain: regulated customers (finance, government) require offline deployment, domestic OS, security grading and audit trails.
With Pigsty:
- Supports RockyLinux, AlmaLinux, openEuler, UOS, Kylin and more (OS compatibility)
- Complete offline installer — delivery without public internet (offline install)
- TLS, password policy, row-level security and audit logging on by default (security hardening)
- AGPL 3.0 license; commercial subscription available
Further reading: Offline install → Security hardening → OS compatibility
🚀 Scenario 6 — SaaS / application templates
Pain: when shipping an application, you want "batteries-included" data layer instead of reinventing the wheel.
With Pigsty: 28+ application templates cover common data apps; one-command Docker Compose:
- BI / dashboards: Supabase, Metabase, Superset
- Low-code / business: Odoo, NocoDB, Directus
- AI / LLM: Dify, Langfuse, Mastra
- Dev tools: Gitea, Harbor, Keycloak
Further reading: Application templates → Docker module
When Pigsty is not a fit
To save you from stepping on mines, Pigsty is not:
- A managed cloud database SaaS — it runs on your machines; there is no hosted service
- A Kubernetes operator — Pigsty targets long-lived "pet" nodes, not ephemeral container scheduling
- A thin PostgreSQL install script — its value is in the full infrastructure, not just the database
If you want "PostgreSQL on K8s", look at CloudNativePG and similar projects.