
| |
|
Minutes of the IAB meeting at IETF 60 were approved.
IESG
Rob Austein, Bert Wijnen:
RFC Editor
Joyce Reynolds:
IRTF
Vern Paxson:
ISOC
Lynn St Amour:
ICANN - Liaison to the Board
John Klensin:
RSSAC
Rob Austein:
IEEE 802
Bernard Aboba
The IAB considered the process to be used in appointing a chair of the IRTF. Further consideration of the process in relation to the role of the IRTF will be continued by email discussion.
The IAB was briefed on the current status of IAB requests to the IANA. IAB concerns in this matter will be pass to ICANN.
A progress report submitted to the IAB is attached.
Response to ICANN request relating to any IAB coment regarding the .net registry function
[Leslie Daigle and Geoff Huston recused themselves from IAB consideration of this item due to potential conflict of interest considerations]
The IAB will pass a public response to ICANN, referring to existing IETF documents and current and poteential DNS-related Working Group activities relating to operation of top level domains.
Date: Mon, 13 Sep 2004
From: Thomas Narten
To: Leslie Daigle
The following is a status update from the IAB ad-hoc committee that is looking at IPv6 addressing issues.
First, the following is what we believe our work items are.
clarifying the documentation (in RFCs) about how the IETF wants IANA to manage the IPv6 space. e.g., IANA allocation sizes to RIRs:
HD ratio. Some questions are being raised as to whether the HD ratio is really the proper metric for address utilization and allocation. Question: Is the current HD ratio model (and policies based on them) still the best model, or should we be considering adjustments?
IANA allocation page cleanup (needs update to reflect reality, still uses TLA field which has been deprecated in RFC 3513, consistent and clear terminology for allocation state)
ip6.int deprecation. Now that we have ip6.arpa, can we stop using ip6.int? RIRs are currently populating it, and would like to stop doing so.
Status: we believe an ID needs to be published and reviewed by the community, then pushed through the system as, e.g., a BCP. Key issue is deciding a date for when deprecating ip6.int makes sense, based on what current implementations actually do.
To date, we've had email exchanges and one conference call. Some early thinking (not a formal recommendation) at this point, includes:
We believe the IETF has a role to play in stewardship of the IPv6 address space. While it is clearly not appropriate for the IETF to micro-manage the IANA/RIR process, the IETF has a long-term interest in seeing that the IPv6 address space is managed prudently, and that a proper balance between aggregation and address conservation is achieved, as envisioned by the IPv6 architecture. In particular, the emphasis on aggregation should be significantly more pronounced in IPv6 than it is (or has been) for IPv4.
IANA appears to be using obsoleted instructions and thinking. See http://www.iana.org/assignments/ipv6-tla-assignments. The TLA concept was obsoleted by RFC 3587 (there is no mention of this RFC on the web site). We will work on cleaning this up.
IANA continues to assign /23s to RIRs (i.e., as recently as September, 2004). This is clearly too small a size, and reflects a thinking that pre-dates both RFC 3177, and the global IPv6 policy that the RIRs are currently using. For long-term support of aggregation, it is important that ISPs be allocated sufficiently large chunks, plus, RIRs must "hold back" sufficient space "in reserve" to allow an ISP's allocation to grow in the future in such a way that its existing allocation can be expanded into a shorter prefix, rather than requiring that a new, separate allocation be made (i.e., fragementation of the address space).
An important goal for IPv6 is to support good aggregation over long time periods (e.g., 10-20 years and beyond), so that as an ISP grows, it continues to be able to use a single prefix to aggregate all of its customers.
We are currently studying what a more appropriate IANA-RIR allocation size is; initial discussions have suggested something in the range of /12-/16 being appropriate.
The topic of IANA->RIR allocation sizes is being actively discussed within the RIR community. The expectation/plan is that each RIR will discuss the issue within its community with the eventual goal of having a single global policy that can be submitted to ICANN through the ASO. Note: it goes without saying that any such eventual recommendation should be in sync with the IETF/IAB view, so it is important that we follow/engage in these discussions as they happen.
References:
RIPE 49, Manchester, 20-24 September:
http://www.ripe.net/ripe/draft-documents/ipv6.html
ARIN XIV, Reston, 20-22 Oct:
http://www.arin.net/mailing_lists/ppml/2758.html
LACNIC VII, San Jose (Costa Rica), 26-28 Oct
http://lacnic.net/en/meetings.html
APNIC 19, Kyoto, 22-25 Feb 2005
http://www.apnic.net/meetings/19
The HD Ratio is currently used to determine when an ISP has sufficiently used up a current allocation and is eligble for more address space from an RIR. (Specifically, an HD ratio of 0.8 is the current threshold.)
At present, there is no formally defined/accepted definition of when an RIR has sufficiently "used up" an allocation it has received from IANA and becomes eligible to get more. This is a critically important topic, and the same HD ratio target is not appropriate for this. Indeed, it is unclear that the HD ratio is even an appropriate metric in this context. The issue is that in order to ensure good aggregation in the long term, RIRs must also leave their space "underutilized" or "sparsely allocated", in order to ensure that as an ISP grows, it can get adjacent address space to its already allocated block. This emphasis on aggregation in IPv6 contrasts with the current practice in IPv4, where emphasis is weighted more heavily on address conservation.
Further work/discussions are needed to better understand what would be good/acceptable policies/practices here. Geoff has been doing some modeling/simulation work in this space, based on data from actual IPv4 customer allocation history.
These minutes were prepared by Geoff Huston; comments should be sent to iab-execd@iab.org. An online copy of these and other minutes is available at: http://www.iab.org/documents/iabmins/
The IAB Web page is at http://www.iab.org
|
![]() Index of IAB Meeting Minutes |
|