We build telemetry software for workstation fleets, virtualized infrastructure, embedded devices, and other systems where collecting diagnostics must not become the thing causing the problem.
FleetWatch collects telemetry from computers. ChipWatch records structured events in embedded systems. Neither attempts to “discover everything.”
A lightweight telemetry agent and collection system for engineering workstations, virtual machines, render nodes, laboratory systems, and long-lived servers.
A C/C++ telemetry library for constrained firmware. It can operate without dynamic allocation, threads, filesystem access, or an IP network.
This was not originally an architectural statement. Erlang was simply the thing that kept working. C and C++ remain where predictability and hardware access matter.
When resources become constrained, collection becomes less detailed before the host becomes less reliable.
Queues, buffers, retries, and logs have limits. “Effectively unlimited” is only a missing maximum size.
A clever distributed system is entertaining exactly once. A boring one can be operated for ten years.
For regulated, offline, or otherwise sensitive installations, Hardened Builds may include reduced diagnostics, stripped metadata, customer-specific feature sets, signed offline updates, and additional binary hardening.
Hardened Builds and securityMost of our staff write software, operate infrastructure, review protocols, or answer technical support requests. The engineer answering your ticket probably wrote the code involved.
01 -module(collector).
02 -behaviour(gen_server).
03
04 handle_cast({event, Signal}, State) ->
05 ok = pipeline:ingest(Signal),
06 {noreply, State};
07
08 handle_info({'DOWN', _, process, _, _}, State) ->
09 {noreply, recover(State)}.Tell us about the fleet, the constraints, and the signals that matter.
sales@insight-technology.org