Self-hosting Postgres is cheap. Until it isn’t.
The server costs a few euros a month. Backups, restore testing, patching, capacity planning and being on call cost your engineers’ time — every week, indefinitely. Balta takes that work off your plate for one flat monthly price.
The server is the cheap part.
Every self-managed database comes with five ongoing jobs. None of them is hard. All of them quietly stop being done the day the person who owned them moves on.
| The job | What it takes | When nobody owns it | With Balta |
|---|---|---|---|
| Backups | Configure archiving, monitor it, watch for gaps, manage retention. | The job succeeds every night and nobody has opened a file. | Continuous archiving to two repositories, monitored for you. |
| Restore testing | Restore a real backup on a schedule and check it starts. | The first restore happens during the incident. | Automatic, on a schedule, with results in your dashboard. |
| Patching | Track security releases and apply them safely. | Two major versions behind, and the upgrade is always next quarter. | Minor updates applied for you in your maintenance window. |
| Capacity | Watch storage, connections and I/O before they run out. | The disk fills on a Sunday. | Dedicated resources, visible in your dashboard, storage expandable online. |
| Monitoring | Know something is wrong before your users do. | The first alert is a customer email. | Every service monitored around the clock. |
How they compare in production.
| Self-managed | Balta DB | |
|---|---|---|
| Total cost | A cheap server, plus the engineering hours for backups, restore testing, patching and monitoring — every month. | One flat monthly price that includes all of it. |
| Backup verification | You build it — and most teams never test a restore until they need one. | Automatic restore tests, with results and timings in your dashboard. |
| Recovery | Depends on who’s available and whether the runbook still works. | Rebuilt from a verified backup onto reserved capacity, at the same connection address. |
| Recovery point | Whenever your last successful archive happened to be. | Five minutes or better, monitored continuously. |
| Security & audit | Your responsibility, from OS hardening to audit trails. | Per-service isolation, least-privilege access, MFA and a customer-visible audit log. |
| Extensions | Anything — and you own the consequences. | A curated catalogue, including pgvector, pg_trgm and pgcrypto. |
| Everyday tooling | Whatever you install and maintain yourself. | Built-in data browser, SQL editor and protection monitoring. |
| Your data | Yours. | Also yours — standard PostgreSQL, pg_dump any time, one-click export. |
Reserve capacity
Somewhere to recover to, the day you need it.
The hardest thing for a single team to have is spare capacity. Every Balta location keeps enough capacity free to rebuild a failed host’s services on the hosts that remain — and we stop taking new services before we’d ever touch it.
A team running one server has exactly one machine. When it fails, the recovery starts with ordering another one.
Self-hosting works if you have an engineer who enjoys running databases and has time to spare. Balta is for teams whose engineers should be shipping product instead.