Summary

  • Matthew Wilder introduced ARIN-2026-2 as a way to replace formula-heavy IPv6 allocation rules with network planning. The draft asks applicants to document assignment sizes, hierarchy and one-, two- and five-year growth.
  • At ARIN 58, speakers asked why every applicant should do that work. Kevin Blumberg proposed a size distinction around /32; ARIN staff described guidance already available for smaller networks. Wilder left the meeting with a sharper question: where does advice end and mandatory paperwork begin?

Matthew Wilder came to ARIN 58 to make IPv6 allocation easier to understand. His slides were so full of text that he told the audience to open the draft on their own devices. It was an honest opening for a simplification project.

The ARIN-2026-2 proposal would trade much of NRPM Section 6.5’s formula-driven sizing for a documented network plan. An applicant would describe typical subnet sizes, the shape of its address hierarchy and expected use at roughly one, two and five years. The plan would help staff judge an initial allocation and later requests. The aim was to let network design, aggregation and room to grow do more work than arithmetic written for an older era.

Wilder, the policy shepherd, did not pretend the swap was settled. He told the room he was unsure whether the community supported requiring a plan. A survey respondent had asked whether ARIN would provide a template. At the microphones, the questions became less polite and more useful. Kevin Blumberg said small allocations should come with practical guidance, not compulsory plans; a request as large as /16 could justify heavier scrutiny. Eric Landgraf saw another form to review and no clear cure for deployment complexity. Both suggested the burden should rise with the size of the ask.

ARIN’s John Sweeting described an existing recorded guide that asks about a network and recommends sizing from its largest serving site. That guidance, he said, works through a range of smaller allocations; very large requests need a plan. He was describing practice, not announcing a new threshold. Wilder brought his own operating experience into the exchange: his organization had a /32 with space reserved out to /29 and later obtained the rest. He also questioned the old insistence on neat four-bit boundaries for large multi-region networks.

One early IPv6 holder offered a less tidy story. Its original block had been in use for more than 27 years; neighboring allocations now constrained contiguous growth. The request was for guidance on how much space to give each site, including dams, without painting the organization into another corner. Wilder agreed that guidance mattered. The unresolved placement of that guidance was the point: in mandatory policy text, or in a tool people will actually consult before they file?

The draft remains under discussion. Wilder's simpler rule has to survive a familiar test: an application for a modest network should not require the sort of map a planet-wide operator would bring to a merger.