My take (albeit watching from the sidelines here myself):

While the number of /64s available in a /32 makes this largely theoretical, I’d 
expect that if an operator were to dedicate /64s for this purpose, they would 
not be able to count them towards their HID ratio for a subsequent allocation.

An operator can certainly use a subset of a /64 provisioned for 
“active-networking" purposes to number non-network-active IoT devices, as long 
as they don’t collide with other addressing including stateless EUI-64 ranges. 
These would presumably be able to be counted towards the HID ratio.

This does not solve the issue I raised earlier that IP resources are generally 
intended to be ephemeral and location/topology-specific, and that if this type 
of addressing permanently associates any given “thing” with an IP address, I’d 
argue that raises a substantial risk of larger issues down the road.

Thanks,

-Chris

> On Aug 19, 2024, at 08:42, Rob Evans <[email protected]> wrote:
> 
> Hi,
>> However, that agreement may not be necessary. The holder of a /32 allocated 
>> under current policies can use that space legitimately, while also having 
>> plenty of spare /64s. And each of those /64 prefixes bears enough /128 
>> addresses to identify 2^64 (18 billion billion) objects.
>> 
>> Perhaps the authors might consider that if an existing IPv6 holder wants to 
>> implement their scheme, they could simply take a single /64 from their IPv6 
>> space and use it as proposed. That could even be the same /64 which is 
>> assigned to the server of information about the objects identified within 
>> the prefix (an implementation detail which is probably beyond the scope of 
>> this discussion).
>> 
> 
> Very much this. I must confess that I’ve only been peripherally watching this 
> discussion and had meant to go back and read it in more detail to try and 
> understand why this wasn’t an option…
> 
> Cheers,
> Rob
> 
> _______________________________________________
> SIG-policy - https://mailman.apnic.net/[email protected]/
> To unsubscribe send an email to [email protected]

_______________________________________________
SIG-policy - https://mailman.apnic.net/[email protected]/
To unsubscribe send an email to [email protected]

Reply via email to