>On Thu, May 24, 2018 at 7:10 AM James Morrow via ipv6-wg <ipv6-wg@ripe.net>
>wrote:
>
>> i read this document with a mixture of astonishment, confusion and
>horror.  it's awful.
>
>> the document is utterly, utterly broken.  it has no redeeming or
>worthwhile qualities at all.
>
>> the only thing it's good for is an example of how #not# to do an ip
>addressing plan based on errors and poorly articulated, mistaken
>assumptions.  it also shows beyond any doubt that itu should not meddle in
>ip addressing because it has no competence or mandate to get involved in
>this field.  i've already seen far too many deeply flawed itu documents on
>ipv6, such as the 2009? nav6 study.  this one's much, much worse.
>
>> it's painfully obvious whoever wrote this junk has no understanding and
>operational experience of how to design or deploy an ipv6 addressing plan
>for any non-trivial ipv6 network.  the document is not a sound piece of
>work that makes any sort of technical or engineering sense.
>
>> the document is riddled with errors - far too many to list.  it makes
>ridiculous assertions that have no basis in fact and does not provide any
>references to justify these claims or let someone check them.  i started to
>write down these flaws and then gave up in disgust.  why should i do
>somebody else's homework for them?  conflating ipv6 uptake rates with
>developed/developing countries is yet another serious failing.  these
>things are completely orthogonal to each other.
>
>> the unstated assumptions are wrong too.
>
>> first of all, the notion of "special" ipv6 addressing plans for iot
>devices is foolish.  these don't need to be treated differently and
>shouldn't be treated differently from anything else that's connected to the
>internet - at least from an addressing perspective.
>
>> next, it's beyond absurd to suggest or imply there could be one
>over-arching addressing plan that can be used and will work perfectly for
>iot devices in any network or every use case.  that's just basic common
>sense.  how you'd do that depends on the actual network and its
>requirements.  for example take smart lightbulbs: an addressing plan for
>home use wouldn't be suitable for a large building (school, hospital,
>office block, etc) or for a town's street lights.  they'd all have
>different (subnet) addressing plans that were suited to their specific
>needs - number of lights, topology, security, planned expansion,
>architecture(s), link-layer connectivity, redundancy / spofs, budget,
>latency, bandwidth, interoperability and compatibility with existing
>systems / networks (if any), access controls and so on.  the document
>doesn't even hint at any of those considerations.
>
>> the only thing to be done with this document is kill it.  kill it with
>fire.  it's too far gone to be fixed or salvaged..
>
>> imo, the wg needs to tell itu to stay well away from ip addressing and
>leave this to the experts who actually build and run ip networks.
>