Summary

  • APNIC lists prop-170 for discussion at the APNIC 62 Open Policy Meeting; it was not adopted or implemented at the reporting freeze.
  • The proposal would preserve needs-based assessment while allowing the smallest nibble-aligned IPv6 allocation that can satisfy a validated requirement.
  • The governance question is not whether hexadecimal boundaries are convenient, but whether the address-space increment created only by alignment remains explicit, reviewable and conditional.

IPv6 can route a /29 as readily as a /28. People and operational systems, however, often find the second block easier to read and divide because its boundary coincides with a hexadecimal digit. That difference between technical validity and operational legibility is the opening for APNIC prop-170.

APNIC records version 1 as posted to the Policy SIG mailing list on 25 July 2026. It lists the proposal for discussion at the APNIC 62 Open Policy Meeting and records a Secretariat impact assessment posted on 18 August. Those are stages in a policy process, not evidence of consensus, endorsement or implementation.

The proposal applies to IPv6 allocations for Local Internet Registries under Section 8. It does not amend IPv6 assignments under Section 9. A nibble boundary is a prefix length divisible by four, such as /20, /24, /28 or /32. Because each hexadecimal character represents four bits, such blocks can map more cleanly into address plans, automation, security zones, documentation and reverse-DNS structures.

Need first, alignment second

Prop-170 does not propose a free-standing right to a larger block. For an initial allocation, an applicant would have to request nibble alignment and provide a long-term IPv6 address plan. APNIC would first calculate the requirement under its existing needs-based rules, then normally delegate the smallest nibble-aligned block capable of satisfying it.

The order matters. If a calculated need can be met by a /30, the proposal's own example says the smallest satisfying nibble-aligned block would be a /28. The two-bit difference is not additional demonstrated demand. It is the space cost of the requested alignment.

For subsequent allocations, the existing one-bit doubling mechanism would remain the default when eligibility rests on acceptable utilisation. A holder seeking more than that default would still have to justify and validate the larger requirement. Only then could APNIC expand the existing allocation to an appropriate nibble boundary where possible, or make a separate justified allocation where adequate contiguous expansion is unavailable.

The draft also preserves the /32 minimum, HD-Ratio utilisation requirements, documentation rules, the Sparse Allocation Framework and the exception for multiple discrete networks. Existing holders would not be required to renumber, return or resize their current prefixes merely because the policy changed.

These exclusions keep the proposal narrow. They do not, by themselves, show how consistently the new discretion would be exercised.

Make the alignment increment observable

The proposers argue that nibble-aligned blocks are easier to read, document, subdivide, automate and audit, and that alignment can reduce repeated incremental requests, fragmentation and future renumbering. These are plausible operational benefits. The source package does not quantify them.

It also does not say how many APNIC or NIR requests produce non-aligned calculated requirements, how much additional space alignment would delegate, how often adjacent expansion would be impossible, or how much administrative work would be avoided. The absence of those numbers does not invalidate the proposal. It identifies what an implementation should measure.

Every aligned decision can preserve two values: the validated non-aligned requirement and the final delegated prefix. The difference is the alignment increment. Recording that increment, the planning evidence relied upon and whether the result was contiguous would let APNIC and the community test the policy's actual trade-off without exposing confidential network plans.

Aggregate reporting could answer practical questions. How often was alignment requested? How often was it granted? What fraction of aligned decisions required a separate block? How much address space was added solely to reach the next nibble boundary? Did aligned holders make fewer incremental requests or avoid later renumbering?

Without that record, two unlike judgements could look identical. One reviewer might treat a long-term hierarchy as adequate evidence; another might expect detailed customer, site and segmentation forecasts. NIR and APNIC processes could drift while every decision remained nominally “needs based”. A small decision ledger would make discretion comparable without turning a flexible planning mechanism into a rigid formula.

Interaction with prop-164 needs a visible resolution

APNIC's Secretariat says registry tools, internal guidance and assessment processes would need updating. It identifies no material legal issue and estimates approximately six months for implementation if the proposal reached consensus, subject to editorial comments. It also notes that drafting revisions would be required if prop-170 and prop-164 both reached consensus.

That interaction is substantive. Prop-164 concerns IPv6 allocations longer than /32 with nibble-boundary alignment, while prop-170 preserves the current /32 minimum and adds conditional alignment for larger initial and subsequent LIR allocations. The final policy text would need to make the scopes and sizing sequence unambiguous. A collision resolved only in internal guidance would leave applicants unable to tell which rule governs a request.

The strongest version of prop-170 would therefore state the decision sequence in public policy and report its effects: eligibility, calculated need, explicit alignment request, supporting long-term plan, smallest satisfying nibble boundary, adjacency result and alignment increment. That sequence protects both sides of the bargain. Operators gain a route to cleaner hexadecimal planning. The registry retains evidence that operational convenience did not silently become an entitlement to unmeasured space.

Sources