Summary

  • A 27 October revision of an IETF v6ops working-group draft adds a concrete switching case: an end site planned around a /48 may receive only a /56 from the next provider. The old internal subnet layout then stops fitting.
  • The same draft recommends a persistent /48 for end sites and proposes /56 as the longest acceptable prefix. The choice between /48 and /56 is an operator product decision whose migration cost lands on the customer.

An ISP can call a smaller IPv6 block a residential tier. The customer calls it a weekend with the network diagram. In revision 01 of the v6ops prefix-assignment draft, published on 27 October, a new paragraph puts two providers side by side. One delegates a /48; the next offers a /56. A network laid out under the first size may need its subnet plan rebuilt for the second. The same mismatch can appear when two links are used for backup or multihoming.

Those eight bits are a commercial decision with an engineering invoice. A /48 contains 65,536 possible /64 subnets; a /56 contains 256. That arithmetic does not say a small business uses thousands of networks. It says the smaller block gives a different planning canvas. A customer who has placed services, guest networks and equipment across that canvas cannot treat a change of provider as a mere cable swap.

The draft's older revision 00 already discussed business /48 and residential /52 or /56 tiers. Section 4.2 says sales and marketing often drive that distinction, even though small businesses may buy residential service. The October revision sharpens the consequence across providers: if the sizes differ, the customer's plan may not travel. A matching prefix length can preserve the shape of the internal layout while the provider-assigned address values change. It does not make those addresses portable.

The proposed Best Current Practice aims to make /48 and persistent delegation the ordinary offer. Its summary also sets a floor against giving end sites anything longer than /56. At this stage it is a working-group draft; the text, its status and any operator adoption remain open decisions. What is already visible is the cost allocation: providers choose the tier; customers do the renumbering when a later tier is narrower.

The first useful test is a provider's actual product sheet beside a migration plan. Which prefix size is promised, does it persist, and what happens to a customer with a /48 plan who moves to a /56 service? Without those records, “easy to switch” is a brochure line, not an engineering result.