Twenty-seven years of watching networks, turned into something you can host.

Where this comes from

We have built most layers of this stack before, for someone else.

Our background spans the full evolution of modern networking, and each stage left us with an opinion about what the next tool should do differently.

Protocol

RFC conformance and protocol validationTesting implementations against the specification, which is where you learn that vendors interpret the same document differently — and why vendor-aware handlers matter more than generic MIB support.

Routing

Carrier-grade routing platformsBuilding for environments where an outage is measured in customers, not tickets. Scale stops being a benchmark and starts being a design constraint.

Security

IPsec and secure networkingWork that makes the case for self-hosting obvious: some organisations cannot send their topology to a third party, and no pricing discussion changes that.

Packets

Packet capture and deep inspectionThe ground truth layer. Flow records tell you what happened; packets tell you why, and knowing both is what lets a probe emit TCP performance rather than raw counts.

Observability

High-speed telemetry and streaming analyticsReal-time pipelines turning large volumes of network data into something operationally useful — the direct ancestor of what NitroNet does with goflow2, Kafka and ClickHouse.

Why we built it

Monitoring got expensive and complicated at the same time, and neither was necessary.

NitroNet was built for engineers who care about network monitoring but want it at a reasonable price and without unnecessary complexity. That comes from years of direct customer feedback and field deployments — from watching what people actually do at 2am, rather than what the feature matrix says they should.