Home / How it works
How Azimuth works.
Readings come in from your sites, one engine decides what is wrong, and your team hears about it and fixes it. Here is every part, and what happens between a missed ping and a fixed site.
The parts.
Four columns, left to right: your sites, the way in, Azimuth, and your team.
Click any part to see what it does.
How Azimuth fits together
Readings come in from your sites through the controllers, the agent, Azimuth's own checks and site tunnels. One engine decides what is wrong. Your team hears about it, works it, and fixes it.
Click a box on the diagram to see what each part does and what it collects.
Network controllers
AvailableAzimuth reads Omada and UniFi controllers through their own APIs: every gateway, switch and access point, their clients, WAN ports and firmware. Nothing extra is installed at the site.
- Omada: cloud or direct, WAN per port, site-to-site VPN state
- UniFi: Site Manager API key
- A quiet controller is diagnosed: site down, service stopped, or login refused
- Devices it stops listing are kept and flagged, never retired automatically
The Azimuth Agent
AvailableA small program for Windows, Linux, Synology and TrueNAS. It installs with one line, enrols with a code and reports outbound only, so no port is ever opened on the customer's firewall.
- Health: CPU, memory, disks, temperatures, SMART, RAID and ZFS
- Security: Defender, firewall, BitLocker, pending updates
- Inventory: hardware, serial, addresses, installed software
- Watches printers, cameras and switches on its own network
Websites and online services
AvailableWebsites, web apps, certificates and domains are checked from Azimuth itself. Nothing is installed anywhere; you only give the address.
- Uptime and response time, continuously
- Keyword present on the page
- Certificate expiry, domain registration expiry
- DNS resolution and mail records
Azimuth Edge
Coming soonA small box for sites with no server: it plugs into the site network, watches everything on it and holds its own encrypted tunnel to Azimuth.
- One product for offices, factories, hotels and ships
- Collects from switches, cameras, phones, UPS and sensors
- Buffers readings when the link is down or costly
- Coming soon
The WireGuard hub
AvailableEach site connects to the Azimuth hub over its own WireGuard tunnel, started from inside the site. Sites behind CGNAT, Starlink or with overlapping addresses join as they are.
- Outbound from the site: no ports opened
- 1:1 address mapping: customer networks are never renumbered
- Tunnel quality measured every minute: latency, loss, jitter
- Gateway settings ready to copy, shown to admins only and logged
Checks from Azimuth
AvailablePing, website, certificate, port and DNS checks run from Azimuth's own servers, each on its own schedule.
- Each check has its own interval and timeout
- One slow site never delays the others
- Late runs are recorded and shown
The monitoring engine
AvailableEvery reading goes through one engine. It decides what is up, degraded, stale or down, opens and closes incidents, and keeps the history.
- State machine: misses make a device unstable, more make it down, hits bring it back
- Stale, not green: old data turns grey
- Flapping links send one message, not a storm
- Maintenance windows never alert or dent uptime
- Raw readings kept 7 days, hourly figures 400 days
Alerts
AvailableWhen something needs a person, Azimuth tells the right people by severity, and keeps reminding until someone picks it up.
- On screen, by e-mail and by Telegram (push notifications coming)
- Acknowledge, snooze or open a ticket
- Repeat reminders for unacknowledged critical alerts
- Rules per company: flapping limits, reminder interval
The console and the globe
AvailableWhere people work: the Overview command wall, the 3D globe, the wall screen for a TV and the phone layout for technicians.
- Overview: problems first, then customers, risks and jobs
- Globe: sites, vessels and links on a real globe
- Wall screen: read-only, from a link you can revoke
- Phone layout with buttons sized for thumbs
Tickets
AvailableLight, NOC-native tickets: from an alert, the globe or a user's Request support button, with a technician, priority, due date, timeline and field visits.
- Numbers per company, assignee told by e-mail or Telegram
- SLA countdown on the Overview
- Field visits appear on the globe's schedule
The life of an alert.
What happens between the moment a gateway stops answering and the moment it is back, step by step.
A check runs
Every check, controller and agent has its own interval. The Warehouse gateway is pinged every minute.
A miss is not an alert
One failed reading makes the gateway unstable, not down. Networks hiccup; Azimuth waits for proof.
Down, incident opens
After several misses in a row the gateway is down. An incident opens with its cause and a timeline.
The right people are told
Critical goes on screen, by Telegram and by e-mail, following the company's alert rules.
Someone acts
A technician acknowledges, snoozes or opens a ticket in one click. Unacknowledged criticals are sent again every 30 minutes.
Recovery
Successful readings bring the gateway back. The incident closes itself and its timeline stays.
History
Raw readings are kept 7 days and hourly figures 400 days, so uptime for 24 hours, 7 days and 30 days is always to hand.
Try it
Who gets told, and when.
Set up channels and rules the way you would in Azimuth, pick what goes wrong, and see every message that would go out, minute by minute.
Your channels
Every alert is always on screen. Each e-mail address or Telegram chat chooses what it receives.
Your rules
Flapping: 3 drops within 30 minutes send one message, and the alert closes itself after 30 calm minutes.
A simulation of Azimuth's real alert rules. It runs in your browser and nothing is sent anywhere.
Four ways to connect a site.
Use one, or all four. A new company is walked through its first site and first equipment when it signs in for the first time.
For PCs, servers and NAS.
- Create an enrolment code for the site
- Copy the one-line install command
- Run it on the machine: Windows, Linux, Synology or TrueNAS
- The machine appears within a minute
About 2 minutes per machine
For Omada and UniFi networks.
- Add the integration with its API key
- Press Test connection before saving
- Link each controller site to an Azimuth site, or create one
- Every device appears, matched by MAC, never duplicated
About 5 minutes per controller
For websites and web apps.
- Paste the address
- Azimuth fills in uptime, certificate and domain checks
- Add a keyword the page must contain, if you like
Under a minute
For sites behind CGNAT or Starlink.
- Create the tunnel on the site page
- Copy the gateway settings into the gateway
- The tunnel connects outbound; quality is measured from the first handshake
About 10 minutes per site
What is kept, and for how long.
Enough history to answer what happened last month, without keeping every reading forever.
| Raw check readings | 7 days |
| Hourly figures (uptime, latency) | 400 days |
| Incidents and their timelines | Kept |
| Audit log of changes and remote actions | Kept |
| Times | Stored in UTC, shown in your time zone |
See it working on your own sites.
We'll connect one of your sites in a demo, so you can see the engine at work on your own equipment.