some corrections / comments
1 Locally assined ULAs are not unique - the use of the term
"probabilistic uniqueness" is meaningless. What is closer to the mark is
the observation that in most context of overlapping use that involve
only a subset of locally assigned UALS the probability of collision is
mall assuming that each local selection was perfectly random
2.1 ULA-Cs are "assuredly unique" i.e. they do not just offer a greater
level of assurance of uniqueness than locally assigned ULAs - they are
intended to be unique.
2.2 confuses the concept of uniqueness and traceability - what is
missing here is the observation that "uniqueness" is by its very nature
a public declaration and that public declaration is usually carried in
some form of managed registry.
Traceability is a separate concept, and due to the central registry it
could be possible to allow ULA-Cs to be traceable
3.2 appears to not to mention the costs of registry maintenance and
looks only at registration. Indeed, one of the major issues with ULAs
has been the assumption on the part of some that registry services are
cheap to set up and cheap to operate, while others observe registry
operations in other parts of the address space and observe that a
managed accurate registry is neither simple nor cheap to operatoe.
regards,
Geoff
Margaret Wasserman wrote:
FYI --
I wrote this draft to try to capture the major arguments for and against
the definition of ULA-Cs. Please let me know if I've gotten anything
wrong, or if there are any major arguments (in either direction) that
I've missed.
Thanks,
Margaret
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
