Summary

  • David Tatlisu declared RIPE Database Working Group item NWI-22 on 29 October 2026, following Ed Shryane’s proposal to reduce the RIPE-NONAUTH database. Opening a work item did not authorize a deletion job.
  • Shryane’s 9 October problem statement proposes two candidate classes; exact matching, advance notice, maintainer opt-out and publication of the cleanup procedure await a later Solution Definition.

A number for the problem

Tatlisu, a network engineer and Database Working Group co-chair, made a small but consequential move just before RIPE 93. In the group’s public thread, he said participants had had time to express their views and declared the cleanup problem NWI-22. A work-item number gives a proposal a place to be worked on. It is not a deletion order.

The underlying figures are large enough to make that distinction matter. Ed Shryane says RIPE-NONAUTH holds about 44,526 objects and sees fewer than 100 updates a year. He proposes removing objects that match a valid ROA, a candidate class of about 16,497, and objects matching a record in another RIR database, a class of about 27,449. He estimates an initial reduction of roughly 60%. The two class counts cannot simply be added: his public post gives no distinct-object overlap. Nor does it show the final list of records to be removed.

NONAUTH records are nonauthoritative out-of-region routing-registry data. Shryane says they were created anonymously with a shared password, leaving their origin and continued validity uncertain. That makes cleanup attractive. It also makes a sloppy matching rule dangerous. A covering prefix and a more specific route are different objects; one matching ROA or registry entry should not become a vague instruction to erase a neighbour.

The safeguards come after the headline

Shryane’s message reserves four design questions for a later Solution Definition: how to match candidates, how to notify maintainers, how to let them opt out, and how to document the cleanup. He excludes complete retirement of NONAUTH; queries and NRTM mirroring would continue. Tatlisu’s 29 October announcement placed that unfinished design on the group’s work list. The RIPE 93 transcript records him announcing the new item, without supplying a final deletion specification.

The missing evidence is a dated Solution Definition with an exact candidate list, notice window, opt-out receipt and reversal procedure. Until then, the proposal has a target and a chair’s work-item number, but no safe delete key.

Sources: RIPE Database Working Group thread; RIPE 93 Database session transcript; official working-group chair profile.