Raw telemetry in. Auditable business metrics out.
PathMetrics is one lightweight collector — the Anchor — feeding a formula stack that turns per-hop, bidirectional network telemetry into nine scores your NOC, your SRE lead, and your CFO can all read the same way.
Hop-by-hop, in both directions, at the same time.
This is the section for the people who will actually push back on our claims — and we want them to. Here's exactly what's under the hood.
Multi-protocol, ECMP-safe
ICMP → UDP → TCP-SYN fallback per hop. Paris-traceroute-consistent flow-key pinning means ECMP hashing returns the real path — no phantom hops, no guessing which of sixteen load-balanced routes your packet actually took.
Per-hop MTU discovery
DF-bit / ICMP Fragmentation-Needed probing produces the MTU step diagram — exactly where a tunnel or an overlay quietly shrinks your effective packet size, long before an app team notices dropped payloads.
ΔRTT, hop by hop
ΔRTT = median(RTTₕ) − median(RTTₕ₋₁), combined with RTT standard deviation to name the exact link where a queue is building — not just “the network is slow.”
Simultaneous A↔B
Anchor-to-anchor mode runs forward and reverse measurement at the same time, rendered as side-by-side dual graphs with asymmetry deltas called out explicitly.
Loop & blackhole detection
Routing-loop detection, blackhole detection, and ECMP variance scoring run on every measured path, every cycle — not sampled, not spot-checked.
Synthetic + path, fused
iMon synthetic transaction monitoring ships fused with iPath — when a synthetic test fails, the exact hop causing it arrives with the alert, not in a separate tool.
Nine business metrics. Published math behind every one.
Datadog, Dynatrace, and most “health score” tools give you a number with no formula behind it. PathMetrics ships the formula in the product — and, since August 2026, on the bill.
| Metric | Name | Range | Tells you |
|---|---|---|---|
| PRI | Path Reliability Index | 0–100 | Composite end-to-end health of the forward path |
| PSS | Path Stability Score | 0–100 | How much path metrics fluctuate between cycles |
| MRR | MTU Reliability Rating | 0–100 | Whether the path supports expected packet size without fragmentation |
| PCI | Path Congestion Index | 0–1 | Queue build-up, rolled up hop to path |
| TCI | Throughput Capacity Index | 0–1 | Effective pipe size — MTU × (1 − congestion) |
| ARI | Asymmetry Reliability Index | 0–100 | Forward-vs-reverse path quality difference |
| BSS | Bidirectional Stability Score | 0–100 | Whether forward and reverse are equally stable over time |
| TDC | Trust-Weighted Directional Confidence | 0–100 | Confidence in the measurement data itself |
| BPCI | Bidirectional Path Credibility Index | 0–1 | The master score for a bidirectional connection |
All nine roll up into FPHS — the Final Path Health Score, the one number an executive can read. Governed by a Critical Override: if a single dealbreaker metric (like MTU or reliability) hits Critical, FPHS goes Critical regardless of the rest. No mathematical illusions. Weightages, baselines, and thresholds are configurable per customer — the formula is published, not fixed to defaults that don't fit your network.
// latency vs 200ms (ITU-T G.114) · jitter vs 50ms (VoIP/video buffer break) · loss vs 5% (TCP throughput cliff)
Forecasts, not just dashboards.
PathiQ takes the nine metric streams and predicts what happens next — before your users feel it.
Root-Cause Generator
Translates reliability, stability, congestion, and asymmetry signals into a ranked, human-readable cause chain. “This hop, this direction, this reason” — not “something is wrong.”
Bidirectional Degradation Probability
A single probability that the overall bidirectional path will degrade, synthesized from ARI + BSS + TDC via BPCI.
Path Reliability Forecast
Forward-looking PRI projection with confidence bounds, built on PRI history and PSS variance.
Path Degradation Probability
The likelihood a currently healthy path degrades in the next window — drift in PRI, rising PSS variance, increasing congestion.
Congestion Hotspot Forecaster
Predicts where on the path congestion will form next, not just when — tracked at the specific hop, aggregated to PCI.
Anomaly Detection Model
Learns normal-behavior envelopes across PRI / PSS / CI / MRR and directional metrics. Deviations are scored, not guessed.
App Status & SLA-Breach Forecasts
Forecasts whether an application-level SLA is about to breach — before the breach shows up as a ticket.
Which Hop Will Fail First
Ranks every hop on a path by failure risk, so remediation starts at the most likely point of failure, not the first alert that fires.
Geolocation Failure Risk
Flags region- and location-correlated failure patterns across your Anchor fleet, not just single-path anomalies.
Intelligent Alerting
Correlates alerts across metrics and hops before they reach a human — engineered for roughly 60% fewer noise alerts than threshold-only monitoring.
Latency, Jitter & Packet-Drop Trends
Trend models across the raw signal set, feeding the forecasts above instead of sitting in a separate dashboard.
PathiQ ships 15 AI capabilities in total. Every one is included with every Anchor — AI is never a paid add-on.
Predictions are useless if you can't trust them.
CredScore is a 0–100 credibility score attached to every PathiQ prediction, computed by comparing forecasts against actual outcomes. It's the reason PathMetrics can say “AI” to a skeptical audience without losing them.
- → Policy-driven automation gates: auto-remediate only when CredScore ≥ 90
- → Human-in-the-loop routing below your chosen threshold
- → Low scores automatically trigger model retraining
- → Transparent validation logs, engineered to support PCI-DSS / ISO 27001 review
Ticketing, identity, and access — not a separate SKU.
An Anchor isn't just a collector. Every deployment ships with the operations and access-control layer an enterprise NOC actually needs to run it.
Built-in ITSM, plus your stack
Built-in ticketing is FRT/SLA-aware and AI-prioritized — tickets carry the hop, the root cause, and the CredScore. ServiceNow integration is bidirectional and in production today; Jira Service Management, PagerDuty, and OpsGenie are available as scoped integrations.
LeadLense
An NLP mobile/voice app that talks back — ask your observability data a question instead of opening a dashboard. Email notifications route per-user, per-priority alongside it.
One portal, full control
Endpoint, Anchor, and org management from a single portal, with metric-profile administration and API keys for programmatic access. 12-month data retention on Professional, extended on Enterprise.
Bring your own IdP
Native login plus SSO — Microsoft Entra ID, Google Workspace (domain-restricted), SAML 2.0 for any IdP, LDAP/AD. One login page auto-routes users by email domain, with TOTP two-factor as standard.
SCIM 2.0 & RBAC
Automatic user and group lifecycle sync from your IdP via SCIM 2.0. RBAC applies at the org level, the Portal (MSP) level, and down to per-menu CRUD / export / import permissions.
Multi-org, fully audited
Delegated multi-organization access lets partners switch between every org they manage from one login. Lockout policy, IP allow-listing, password expiry, and a read-only audit log cover every grant, denial, and security action.
Included with every Anchor. Never charged for.
Hop count
Every hop, every path — not a sampled subset.
Test frequency
A 1-minute interval costs the same as a 15-minute one.
Bidirectional mode
The flagship capability, standard in every edition.
AI predictions
All 15 PathiQ capabilities, on every Anchor.
Users
Unlimited, in every edition.
Ticketing
Built in, not bolted on with a separate line item.
If a competitor meters it, PathMetrics probably includes it.
Deterministic-first. AI-second. Non-negotiable.
Every score has a published formula. AI is foresight layered on top of deterministic ground truth — never a replacement for it.