Border Gateway Protocol (BGP)

Border Gateway Protocol is how autonomous systems exchange reachability: each network announces the ip-prefix entries it can deliver traffic to, and neighboring systems propagate those announcements with attached AS paths - dozens of collector vantage points capture the result. There is no central registry of internet routing - the protocol's own messages are the source of truth.

There is no central registry of internet routing - 1 distributed protocol carries the truth. The RouteViews BGP Archive & API publishes global BGP RIB dumps and update streams from dozens of collectors, giving researchers point-in-time snapshots of the full routing table plus the churn between snapshots. RIPEstat's Data API, tagged 'bgp' and 'routing', exposes routing history and historical whois that 'reach back years', so a prefix's announcement history can be queried without operating collectors yourself. Both sit in the internet-services-infrastructure slice, where autonomous-system-number lookups are the join key.

Everything downstream - geolocation of routes, hijack detection, CDN placement studies - inherits the viewpoint problem: each collector sees what its peers announce to it.

BGP data is observational, and its gaps are silent. A collector only records prefixes announced to it, so an analysis built on one vantage point misses routes invisible from there - and nothing in the data flags the absence.

For buyers this has 3 practical effects. Hijack detection depends on update streams captured close to the event; sparse collector coverage during an incident produces false negatives. Topology claims are really claims about observed announcements - RouteViews mitigates this by running dozens of collectors, and a vendor who will not say how many vantage points they operate is selling an unquantified view. And historical work has a floor set by archive depth: RIPEstat's routing history reaching back years is useful precisely because most raw update archives are impractical to store yourself. Any procurement question about 'internet routing data' is therefore a question about collectors, retention and viewpoint diversity.

1. Ask how many collectors feed the dataset. RouteViews publishes RIB dumps and update streams from dozens of collectors; vantage-point count bounds what any analysis can claim to see.
2. Distinguish RIB dumps from update streams. Periodic RIB dumps give table snapshots; continuous update streams capture churn and are what hijack detection actually needs.
3. Test historical depth against your window. RIPEstat's routing history and historical whois reach back years - confirm the archive covers your incident dates before committing.

Entries covering the identifiers and structures inside routing data:

• autonomous-system-number
• ip-prefix