Dear Christopher Hawker,
Thanks for your comments and suggestions. Please see my explanations to your 
concerns below.

________________________________
From: Christopher Hawker <[email protected]>
Sent: Tuesday, 6 August 2024 6:01 PM
To: [email protected] <[email protected]>
Subject: [sig-policy] Re: prop-161-v001: Using IPv6 for Internet of Things 
(IoT) -- correct version


There's a few issues issues with this proposal that cause me to object to it:

  1.
The problem statement reads that the IoT industry needs to assign IPv6 address 
space to non-electronic items, which makes routing the IP space to these 
devices impossible. It also would not meet the "demonstrated need" condition 
for resource applications.

I know assigning IPv6 to non-electronic items will break our traditional 
thinking and the normal way of using IPv6 addresses.
However, there are some benefits of using IPv6 addresses as identifiers for 
non-electronic products, like unique, routeable, verifiable and secure. For 
example, a pharmaceutical factory has been allocated a /32 IPv6 and they routed 
that /32 to the Internet. A pack of medicine assigned an IPv6 address. When a 
patient scans the medicine, it looks up that IPv6 address and returns the 
information about the medicine from the pharmaceutical factory. With secure 
routing like RPKI, only the pharmaceutical factory produces that medicine can 
announce that IPv6 address and returns the correct information. If we use other 
numbers as identifiers, they cannot be routed and difficult to verify, but easy 
to fake.
The IPv6 addresses are still routed in this case, just different way of routing 
from different point of view.

  1.
Number 2 under the Proposed Policy Solution states, "An IoT Object will be 
counted as a normal single host while evaluating subsequent allocation size for 
IoT services". The number of addresses in a /64 is 20 digits long. Given that 
there are approximately 4.29 billion /64 subnets within a /32 supernet, 
assigning a /64 to every IoT device and non-electrical item would lead to an 
inefficient utilization of IP address space. This approach may result in a 
significant waste of available IPv6 address resources, regardless of the number 
of /32 prefixes available.

I think non-electrical items and normal electrical devices only need one IPv6 
address which is a /128. A /64 will have 4 billion times 4 billion /128s, I 
don’t expect any company will come back for more than minimum /32 allocation. 
But you never know, just in case someone comes back, this term is to limit the 
allocation size not waste IP addresses.

I didn’t know there is a policy proposal to reduce the minimum allocation size 
from /32 to /36. I think I can change the default initial allocation to minimum 
allocation size without mention /32.

  1.  As a few others have identified, IP space is not designed for globally 
unique identification methods of non-electronic devices. They are designed for 
connectivity and communication between devices. Using IP space for purposes 
which it is not designed should not (in my view) be written into policy.

It would be a good idea to form a working group on IoT-related topics and the 
BoF is an excellent start and even possibly a guideline on the delegation of 
IPv6 space for IoT-related purposes, however I wouldn't consider writing it 
into policy.


Thanks for the suggestion. I have proposed a IoT BoF at the APNIC 58. I will 
consider your suggestions and see if we should update the guidelines instead of 
changing the policy.

Best regards,

Guangliang (Benny)

============



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

Reply via email to