Summary

  • RIPE NCC proposes two ways to identify redundant RIPE-NONAUTH objects: a matching valid ROA or a matching record in another RIR database.
  • The two estimated groups overlap. Staff projects an initial reduction of about 60%, with matching rules and maintainer protections still to be written.

The tempting sum is 16,497 plus 27,449. It lands at 43,946—almost the whole RIPE-NONAUTH database. Yet RIPE NCC expects to remove roughly 60% of its 44,526 objects. The figures in Edward Shryane's Database Working Group proposal describe two candidate groups, not two disjoint piles. Count the same card twice and the apparent bonfire grows very quickly.

One test would catch an object with a matching valid Route Origin Authorisation. The other would catch an object matched in another regional registry's database. The point is provenance: why should an anonymous, lightly maintained copy keep speaking when a stronger source already says the same thing? RIPE NCC says fewer than 100 updates a year reach this non-authoritative collection. Its 23 October member update presents the cleanup as a proposal and invites feedback.

There is a revealing change in the verb “match.” The existing RIPE-731 procedure removes a NONAUTH route object that conflicts with an RPKI ROA after 14 consecutive days of invalidity, with notices along the way. The new idea goes after redundant agreement as well as old contradictions. That transfers weight from a stale copy to the record with better authority; it also makes the exact definition of a match consequential.

Staff has asked the working group for a numbered work item. A later solution definition would set matching criteria, advance notice, a maintainer opt-out and public documentation. Until that exists, no one can turn the two estimates into a precise deletion list. The proposal keeps NONAUTH queries and NRTM mirroring running and excludes objects created under IANA or ICANN action. The ledger may get thinner; the door remains open.