Packetman saysWelcome to the DataStun documentation. This is where we explain our measurement methods — both for the curious and for the engineers who want every number. There are two layers: these public pages describe the methods in plain language with diagrams, and a deeper staff reference holds the precise thresholds. Pick a category below. Security covers how the measurement fabric stays locked down. Performance and Diagnostics cover how we measure throughput, latency, jitter and loss without flooding your link. Sovereignty covers the self-organizing agent mesh that tests your network from the inside, peer to peer. Most people should start with the user guide.Every other network tool floods your link to guess its speed. We send ten packets and read the physics. These pages explain how — the dispersion diagnostics, the self-organizing agent mesh, and the secure fabric that carries it all.
Public docs describe the methods in plain language with diagrams. A deeper staff reference holds every threshold, estimator, and wire-format detail.
Sign up, install an agent, read the dashboard, and act on what you find — fourteen sections covering every feature.
Open the user guide → Glossary of terms →Throughput, latency, jitter, loss, reordering and path asymmetry — measured with a ten-packet sample instead of a multi-hundred-megabyte flood. The methods, the math, and why a tiny test is more honest than a big one.
Packet-pair dispersion, pre-warm before measure, inter-packet-gap analysis, the trust layer that catches buffer-faked speeds, and protocols used as diagnostic lenses.
Read the methods →The product overview: what the mesh sees, the on-demand deep-dive, and the opt-in packet capture decoded centrally — never stored on your endpoint.
Product overview →Your agents test each other directly, peer to peer, without sending your traffic through anyone’s servers. The mesh organizes itself, heals itself, and reaches across NATs and IPv6 the way a video call does.
The connect ladder (LAN → IPv6-direct → NAT hole-punch → relay), multi-candidate racing, the always-warm port, and the two planes that share it.
Read the architecture →The buyer’s view of the n² measurement fabric: what it answers, who it’s for, and where it beats single-point speed tests.
Product page →The thing that measures your network can’t become the thing that exposes it. Agents are outbound-only with zero inbound surface; every probe is authenticated and replay-protected; public test nodes sit behind four independent layers of defense.
The per-tenant key, the wire format, and the four checks every probe must pass before it’s trusted.
Read the auth model →Hash-only evidence (SIG/NSRL/MBZ/VT), published-CVE lookup, the CISA-KEV actively-exploited cross-reference, and the YARA + ClamAV scan with its false-positive discipline.
Read the methods →The wider security story: blocklist enforcement at the endpoint, graded destinations, exposed-service detection.
Security overview →What we collect (metadata only, never content), where it lives, and how to verify it for yourself.
Trust posture →A number out of a speed test tells you nothing about where the problem is. Continuous, comparable measurements between every pair of your sites locate slowness in the path and prove your ISP’s SLA.
SLA validation, stutter diagnosis, return-path comparison over time, and bottleneck attribution.
Performance overview →Why our “speed test” is one preset of a path-investigation engine, not a flood that taxes your link.
See the approach →Install, dashboard, grades, alerts, alert & log export, host diagnostics, account.
Plain-language definitions for every term used across the product and these docs.
Authenticated REST endpoints for agents, telemetry, reputation, and tenancy.