Products
Six products, each built because a diagnosis needed an instrument that did not exist. The commercial option was too heavy, too shallow, or answered a different question than the one in front of me at the time — so the missing piece got written.
They share a position: no cloud dependency, no per-device licensing, and nothing that stops working when the internet does. Each is free to download from its own page, with a checksum for every file, and the page always offers the newest release.
A network that grew by accident hides its problems in the differences between devices: two sites that chose the same range, a community string nobody changed, telnet open by default, a redistribution with no filter. Across thirty devices in three vendors, reading by eye is how the quiet ones get missed.
Most VPN faults are properties of the path, not of the VPN: a path that silently drops large packets, a NAT that maps ports unpredictably, a firewall that passes one protocol and not another, or a mapping forgotten faster than the keepalive. Each is cheap to measure before a design is quoted and expensive to discover after it.
Traefik is configured by editing YAML on the machine, so a route change needs a shell, and a typo in the static file takes the proxy down at the moment the dashboard becomes unreachable.
Multicast fails silently: IGMP snooping, a missing querier, PIM, an RPF check or a firewall can each stop a group with no error anywhere, so the fault gets settled by elimination and by seniority.
Remote support tools fall back to a server their vendor operates whenever a direct path cannot be made, and they end the connection when a path stops cooperating, which is precisely what carrier-grade NAT, inspecting proxies and satellite links do.
Mixed estates contain devices that expose nothing but a ping, and most monitoring either cannot see them or reports the state now and nothing about last week, so an intermittent fault gets argued about rather than diagnosed.