Method & Limits

RPKI Deployment Tracker · back to the reports · BGPKIT Labs

What is counted

The tracker counts what the global RPKI tree published at the moment of each observation. Every number is a count of published objects, not a routing outcome:

Source and cadence

Each observation is one fetch of the Cloudflare RPKI portal JSON export (an rpki-client run over the five RIR trust anchors), stored as a self-contained snapshot by the BGPKIT RPKI tracking pipeline. Observations are scheduled weekly (Mondays) and were collected at an irregular cadence before 2026-09-14; the trend charts plot each observation at its own timestamp, so a visible gap is a collection gap, not missing RPKI data. Snapshot times are shown in UTC.

Windows and deltas

Each report compares one observation against the observation nearest to seven days earlier. When collection gaps make that impossible the nearest earlier observation is used and the actual measured window is printed on the report (for example, the 2026-09-14 report used a 45.5-day window against 2026-07-31). A delta therefore always states its own window; nothing here assumes an evenly spaced series. The same numbers feed the BGPKIT weekly RPKI briefing, so the page and the briefing agree by construction.

Country and region attribution

Countries and continents are not part of the RPKI objects. They come from ASN registry data (RIR delegated statistics and ASN name files) matched on the origin ASN of each ROA or the customer ASN of each ASPA. Limits:

What this cannot establish

Reproducibility

Every report is downloadable as JSON (index and series; /live/reports/YYYY-MM-DD.json per observation, linked from the report history table) and carries its source metadata, snapshot timestamps and measured window. The pipeline runs on BGPKIT infrastructure from the same snapshots that produce the weekly briefing.