How Nivaronix is built, and what it is for
You are considering trusting us to tell you when something breaks. You should know exactly how much of that trust is warranted, and where it runs out. This page is the architecture, stated plainly.
What Nivaronix is for
Nivaronix is a second pair of eyes on the things that break quietly — certificates that expire, domains that lapse, DNS records that change, security headers that get dropped in a deploy. It tells you before your customer does.
It is not your primary uptime SLA system, and we will never sell it as one. If you need contractual availability guarantees with credits attached, you need a different class of vendor, and we would rather tell you that here than after you have paid us.
The architecture
Nivaronix is operated by one person. The architecture reflects that honestly.
Property | How it actually is | What that means for you |
|---|---|---|
Regions | One region, one host. | If that region has an outage, checks pause until it returns. Your data is not lost, but nothing is measured during the gap. |
Monitoring scheduler | A single scheduler process. No automatic failover. | If it stops, checks stop. This is the failure we watch hardest — see below. |
Check locations | Checks run from one location. | A fault that only affects one part of the world may look fine to us. We report what we can see, not more. |
Backups | Nightly, with a worst-case 24 hours of data loss. | In a total database loss you could lose up to a day of check history. Your assets and configuration are restored; recent results may not be. |
Uptime SLA | None. We do not offer one. | There is no credit to claim because we are not promising a number we cannot hold to. |
Status page | Reports measured data only. | If nothing was measured, it says so rather than showing 100%. An uptime figure with no measurements behind it is worse than no figure. |
What we do instead of failover
The worst thing a monitoring company can do is quietly stop monitoring. Not fail loudly — fail silently, so that the absence of alerts reads as good news. We have engineered specifically against that.
The scheduler publishes a heartbeat. If it has not completed a cycle within five intervals, its health endpoint returns a failure code, and an independent third-party checker — running outside our infrastructure, on a different provider — alerts the founder's phone within minutes. Not by email: email is one of the things that might be broken.
We would rather tell you we were down than let you assume we were up.
How uptime is measured
- Uptime is the proportion of completed checks that succeeded, over the window shown. It is a measurement, not an estimate and not a target.
- If no checks ran in a window, our status page shows no figure at all and says why. It will never show 100% on an absence of data — during a scheduler failure that would be precisely backwards.
- Checks run from a single location, so what we report is what one vantage point saw. We do not extrapolate from it.
- We publish our own status on the same infrastructure we sell. We are exposed to our own failures.
How we handle incidents
- Detection before notification. An external watchdog tells us when our own monitoring stops, because internal monitoring cannot be trusted to report its own death.
- Disclosure by default. Incidents affecting checks appear on the status page. An outage you were not told about is a second, worse outage.
- No silent gaps. If we could not measure something, that period is reported as unmeasured rather than backfilled or smoothed over.
- One person answers. There is no support tier to escalate through, which is a genuine limit on response time and a genuine advantage in how directly you are answered.
When this will change
These are commitments with triggers attached, so you can hold us to them rather than take a roadmap on faith.
- A second host, with scheduler failover — At around 60 paying customers.
- Point-in-time recovery, replacing the 24-hour backup window — When revenue funds a managed database.
- Checks from more than one location — When customers are running assets in regions we cannot see properly from one.
If any of this is a dealbreaker for your use case, we would rather you knew now. Tell us what you need and we will give you a straight answer about whether Nivaronix fits.
Nivaronix, operated by Pravin Gyawali