The problem
Most monitoring tools check your services from one place. If that place can't reach you, or if the thing you want to watch lives inside a private network, you're blind. I wanted monitoring that could run anywhere and still report to one pane of glass.
What I built
BoringStatus splits monitoring into two pieces: lightweight agents and a central monitor.
- Independent agents. A small agent runs wherever you need eyes, a VPS, inside a VPC, on a laptop, and probes whatever you point it at: HTTP endpoints, system services, infrastructure health.
- A central monitor. Agents report in continuously; the monitor stores, alerts, and visualises. Lose an agent and the rest keep working.
- Time-series at the core. Every check is a data point, so the storage layer has to handle high write volume and fast range queries.
Technical highlights
The storage choice made the product. TimescaleDB on top of PostgreSQL gave me time-series performance, continuous aggregates, retention policies, fast windowed queries, without giving up SQL or the Postgres ecosystem. Agents speak a compact protocol to the monitor so the system stays cheap to run at scale.
The result
A monitoring stack that watches services from inside the networks they live in, and turns a firehose of checks into answers in milliseconds.