Continuous path-quality monitoring
Pathproof watches the actual capacity and available bandwidth of every network path you care about — continuously, from distributed vantage points, at kilobits per second of probe traffic. When users say “the network is bad,” you'll know how, where, and since when — with numbers that carry their own evidence.
Capacity holds between 918 and 947 Mb/s for all 30 days. Available bandwidth varies 610–780 Mb/s on normal days, drops to 205–260 Mb/s on days 12 through 15 (marked as the incident window), then recovers.
A speed test gives you one number that blends them all together. Packet-dispersion measurement separates what the path could carry from what it can carry right now — so the diagnosis is in the data.
The path's narrow link is smaller than what you're paying for, or a device renegotiated to a lower rate. Capacity is flat — and flat at the wrong number.
reads: capacity < contracted rate, steady
Capacity is fine; available bandwidth sags on a schedule or after a routing change. Someone else's traffic is on your narrow link. Evidence for a peering or QoS conversation.
reads: capacity steady, ADR dips
Loss bursts, reordering, jitter spikes, hop-count changes. The dispersion probes carry a full path-health suite alongside the bandwidth brackets.
reads: loss > 0, IPDV up, route shift
Drop a lightweight agent at each site — offices, DCs, cloud VPCs. Paths between any two agents get the full double-endpoint methodology; anything else can still be covered by service probes or plain ICMP.
Packet pairs and short trains read the time-structure of the path: capacity, an available-bandwidth bracket, loss, jitter, reordering. Kilobits per second, continuously — not a saturating test you can only run at night.
Baselines per path, alerts when capacity steps down or ADR sags below policy, and reports you can attach to an SLA claim — every number carries its method, confidence, and clock provenance.
Tokens, not tiers. One token is one day of monitoring and storage for one path. Monitor three circuits for a month, spend ninety tokens — the bill reads like the measurement log.
Continuous monitoring and storage, one token per path-day. Every plan includes a 90-day archive of full measurement history — raw samples, methods, confidence, provenance.
1 token = 1 path × 1 day · 90-day archive included
Everything in standard, with a 1-year archive and circuit analysis: an expert-reviewed report on your circuit's capacity, congestion pattern, and the evidence behind both.
1-year archive · circuit analysis · same tokens
Bring your own S3. Point Pathproof at your bucket and your measurement history lives in storage you own — for long-term analysis, for as long as you decide. It's your evidence; you hold it.
Pathproof is the commercial service from badnetworks, built on Pathrate, our open-source implementation of peer-reviewed packet-dispersion capacity estimation (Dovrolis, Ramanathan & Moore, IEEE/ACM ToN 2004). The measurement core is BSD-licensed and free, forever — the service adds the fleet, the history, the alerting, and the support.
github.com/Pathrate