[arin-ppml] Global Uniqueness vs Global Reachability

[Randy Bush]

> I was aware of Central ULA, but I don't think it ever really 
> progressed.  While close its not exactly the same thing as I'm 
> talking about; it uses pseudo random assignment like regular 
> ULA.  The draft hand waves the DNS operations issues, and 
> since it uses pseudo random assignment I'm not really sure 
> how your going to make the DNS delegations work.  The draft 
> also envisions, different Registry operations that is typical from 
> an RIR, while I'm sure ARIN or any other RIR could do it.
> 
> As I said, even regular ULA can meet some of the use cases.  
> SixXS has a registry for regular ULA, but again that is not the 
> same thing either, and at least violates the spirit of the RFC 
> 4193. 
> 
> http://www.sixxs.net/tools/grh/ula/
> 
> Anyway, ARIN already does Micro-allocations for Internal 
> Infrastructure.  Which is much more like what I'm talking about 
> than ULA.  A normal Unicast block with a published range that 
> network operators can apply a routing policy to if they are 
> inclined to.  However, what I'm talking about isn't necessarily 
> for internal.  Its just not necessarily intended to be part of the 
> mythical global route table.
> 
> I believe this is a way to have our cake and eat it too.  We can 
> have both a policy that reinforces the routing hierarchy and that 
> provides addresses for those that are willing to exist outside it.

i am always amused that the same people who sing ipv6 is effectively
infinite advocate this ula crap.  have they learned nothing from rfc1918
space messes and difficulties?

this is ipv6.  if someone needs space, you better have an exceedingly
good reason not to give them vanilla unicast ipv6 space.

randy