On Sep 19, 2017, at 1:28 PM, Leif Sawyer <[email protected]> wrote:
The majority of devices no longer register on SLAAC with MAC-bound
addresses.
Technically, this isn???t true.
The majority of devices now register both one or more privacy addresses
_AND_ a MAC-bound address. The MAC-bound address on such devices is not
used as a preferred or primary address for originating sessions, but can be
used (if known by the remote device) as a stable address to connect to
services provided by the host.
Privacy Extensions for Stateless Address Autoconfiguration in IPv6???, which
is codified in RFC 4941
means randomly generated addresses on a rotating basis.
You could disable SLAAC-PE, and get "effectively static" IPv6 - but it's
really not.
I think the better consideration is that when we talk of allocation and/or
assignment, we are talking of the allocation/assignment of network numbers
end not of host-portions to end devices. As such, I don???t think that the
blurring Albert perceives as being created by SLAAC truly exists regardless
of whether it is static or not.
Owen
Leif
*From:* ARIN-PPML [mailto:[email protected]
<[email protected]>] *On Behalf Of *[email protected]
*Sent:* Tuesday, September 19, 2017 3:25 AM
*To:* [email protected]
*Subject:* Re: [arin-ppml] Revised - Draft Policy ARIN-2017-5: Improved
IPv6 Registration Requirements
[External Email]
Placing ISP/LIR in place of ISP might be the best way to avoid confusion.
As has been pointed out, they are really one and the same.
Otherwise, I think that everything else about the draft is good and
support.
One thing to consider for future discussion is that because of the nature
of IPv6, and its end-to-end nature, and assignment of public addresses,
that the difference between allocate and assign using IPv6 on a specific
/64 segment used for public wifi or otherwise is becoming more fluid.
With SLAAC, an address is formed in part using a MAC address, which
according to the rules for MAC addresses is supposed to be unique. It
could be argued that these addresses are in effect "static", which could
be argued is an assignment of part of the host network's /64, in effect a
static /128 of that network. Due to the rules of SLAAC this happens
without involvement of the host network, other than router advertisements,
since the MAC originates from the guest device, as a different device will
have a different MAC address.
The requirement of at least a /64 in the proposed 6.5.5.4 is good in that
end user networks that have SLAAC cannot be required to register the /128
associated with someones MAC address on their request. Since this limit
is in the proposal, I think we do not need to address the fact that end
user networks running IPv6 and SLAAC in effect are assigning addresses to
end user devices, even though they are not supposed to do this unless the
addresses were allocated to them like an ISP/LIR. Unlike DHCP6, which has
a time limit, one could argue that SLAAC addresses are static.
Something to think about.
Albert Erdmann
Network Administrator
Paradise On Line Inc.
On Mon, 18 Sep 2017, Owen DeLong wrote:
I refer you to section 6.5.1???????
6.5.1. Terminology
The terms ISP and LIR are used interchangeably in this document and any
use of either term shall be construed to include both meanings.
The term nibble boundary shall mean a network mask which aligns on a
4-bit boundary (in slash notation, /n, where n is evenly divisible by 4,
allowing unit quantities of X such that 2^n=X where n is evenly divisible
by 4, such as 16, 256, 4096, etc.)
While it is a little unusual to have definitions outside of section 2,
these were placed here in section 6.5.1 in order to avoid potential
conflicts with certain language that was in section 4 at the time of
writing.
Owen
On Sep 18, 2017, at 1:14 PM, John Santos <[email protected]> wrote:
On 9/18/2017 10:37 AM, ARIN wrote:
The following has been revised:
* Draft Policy ARIN-2017-5: Improved IPv6 Registration Requirements
[snip]
4) Add new section 6.5.5.4 "Registration Requested by Recipient" of
the NRPM, to read: "If the downstream recipient of a static assignment of
/64 or more addresses requests publishing of that assignment in ARIN's
registration database, the ISP should register that assignment as described
in section 6.5.5.1."
I have been under the impression that a common goal of most people
proposing NRPM changes is to eliminate the use of the term "ISP", since it
is not defined in the policy and most or all the relevant sections also
apply to other organizations that, while they re-allocate or reassign
address space, are not, properly speaking, ISPs. Shouldn't this says "LIR"
or "provider" or some other more generic term?
[snip]
--
John Santos
Evans Griffiths & Hart, Inc.
781-861-0670 ext 539 <(781)%20861-0670>
_______________________________________________
PPML
You are receiving this message because you are subscribed to
the ARIN Public Policy Mailing List ([email protected]).
Unsubscribe or manage your mailing list subscription at:
http://lists.arin.net/mailman/listinfo/arin-ppml
Please contact [email protected] if you experience any issues.
_______________________________________________
PPML
You are receiving this message because you are subscribed to
the ARIN Public Policy Mailing List ([email protected]).
Unsubscribe or manage your mailing list subscription at:
http://lists.arin.net/mailman/listinfo/arin-ppml
Please contact [email protected] if you experience any issues.
_______________________________________________
PPML
You are receiving this message because you are subscribed to
the ARIN Public Policy Mailing List ([email protected]).
Unsubscribe or manage your mailing list subscription at:
http://lists.arin.net/mailman/listinfo/arin-ppml
Please contact [email protected] if you experience any issues.
