Summary

  • RIPE proposal 2024-01 was withdrawn in September 2026 after feedback. It would have issued justified IPv6 Provider Independent assignments at four-bit steps such as /48, /44 and /40.
  • RIPE NCC’s impact analysis foresaw more address consumption and short-term work, no expected reduction in the global routing table, and requirements it could not efficiently verify or enforce.

A network with two sites can need more than one /48. The withdrawn RIPE proposal offered an elegant-looking answer: give it a /44, a block containing sixteen /48s, and reserve room for the next step. Then the registry had to explain who would check what happened inside the empty fifteen. RIPE NCC’s own impact analysis turned a tidy bit-boundary diagram into a messy operating bill.

Proposal 2024-01 aimed to ease the work of operators using Provider Independent IPv6 addresses. Its third version would have made justified assignments in nibble steps: /48, then /44, /40 and /36. Two sites could justify a /44; seventeen could justify a /40. The proposal also recommended that RIPE NCC reserve contiguous space up to the next boundary. This was a proposal, never an adopted change. The archive records withdrawal on 23 September 2026 after feedback.

The arithmetic was only the first objection. Under the live ripe-738 policy, assignments are for documented specific purposes and PI space cannot be sub-assigned to another organisation. The draft would have allowed some connectivity to separate entities at the same geographical end site and asked holders to justify larger requests with a 24-month plan. RIPE NCC said parts of that package would be hard to verify, monitor or enforce. What counts as the same site: the building, address, city or country? What happens if the projected use never appears?

A registry that cannot answer those questions cannot turn a reservation into a reliable record.

RIPE NCC estimated high operational impact: software changes, process work, documentation and training. It anticipated higher IPv6 address consumption, more short-term tickets and no reduction in the global routing table, since operators could still route more-specific prefixes. Its analysis supplied the awkward comparison: one holder with two end sites and another with sixteen could both receive a /44; the seventeenth site would qualify for a /40. The Executive Board strongly advised the proposers to revise the text.

The withdrawal leaves the current policy in place. The practical problem remains too: multi-site operators still need a route to grow, while the registry needs rules it can administer without pretending to see every future device or routing choice.

Sources