Solution · on-premise · your own perimeter · full data control

Self-hosted infrastructure monitoring

Run a complete monitoring stack inside your own perimeter — metrics, alerts, dashboards and the audit log all stay on your infrastructure. No telemetry leaving the building, no bill that balloons with cardinality.

The problem

Cloud monitoring means your telemetry — and often PII fragments hiding in labels — leaves for someone else’s servers, while the bill balloons with cardinality and volume. Security and compliance teams push back, yet the classic self-hosted stack of a dozen open-source parts becomes a job of its own.

How Unimoni solves it

Unimoni installs as a single Go binary next to Postgres and a Prometheus-compatible TSDB. It comes up in an hour via docker-compose, sips resources and needs no dedicated team to babysit it. You pick the storage region, keep data in-house and upgrade on your own schedule.

What’s included

  • One 12 MB binary — HTTP cabinet, mTLS ingest and evaluator in a single process
  • Data physically never leaves your perimeter: you choose where it lives
  • No licensing math of “hosts × cardinality × samples”
  • Standard export via Prometheus remote-write — zero vendor lock-in

Frequently asked

How much does a self-hosted install need?+

A single VPS is enough to start: the agent uses ~12 MB and ~30 MB RAM per host, and the core with Postgres and the TSDB runs comfortably on 2 vCPU / 4 GB. Components scale out horizontally under load.

Does data really never leave?+

Correct. In self-hosted mode the entire pipeline — ingest, storage, rule evaluation and dashboard rendering — runs inside your perimeter. Nothing leaves except outbound notifications to channels you configure yourself.

How does self-hosted differ from managed feature-wise?+

It is the same platform. Self-hosted means you own the infrastructure and upgrades; managed means we do. Audit log, SSO, embed widgets and multi-region are available in both.