Margaret Wasserman - le (m/j/a) 3/17/09 1:13 AM:
Hi Remi,
I have a few high-level comments/questions on this draft from my first
reading, and I may have more after I have reviewed it in more detail.
(1) You have indicated that you would like to discuss this draft in
the 6AI BOF, but you have not cc:ed the mailing list for the 6AI BOF
([email protected]). Also, have you talked to the chairs of the 6AI BOF
(Bob Hinden and Dan Wing) about whether they are willing to include
this draft on the agenda, despite the fact that it has not been posted
to the I-D archive? There doesn't appear to be an agenda online for
the 6AI BOF yet, so I am not sure if it will be included.
I started with the NAT66 mailing list, and only followed with WG lists.
See also Dan's answer.
(2) The end hosts in the SAM system need to know their globally
routable addresses, so how can SAM be said to provide address
independence?
Routing within the site is based on local addresses, e.g. ULAs.
That's only in SAM-capable CPEs and in SAM-capable hosts that, knowing
SAM parameters, global-address packets can be encapsulated in
local-address packets.
(3) Exactly what formulation of the end-to-end principle are you
referring to in this paper when you indicate that SAM preserves it in
IPv6?
Thanks for the remark.
There should be a reference, e.g. to RFC 1958.
What is meant is IP-layer network transparency, e2e.
Addresses and ports that are seen by two communicating applications must
be the same at both ends.
My understanding of the end-to-end principle is that it has to do with
putting intelligence at the edges of the network (in hosts vs.
routers/middleboxes) and with putting certain function at the top of
the protocol stack (apps layer vs. lower layers). This is based on my
understanding (and recollection) of a paper by Jerry Saltzer, D. Reed
and Dave Clark written in the mid 1980s, which you can find here:
http://web.mit.edu/Saltzer/www/publications/endtoend/endtoend.txt
It is also reasonably well-summarized in this Wikipedia article.
http://en.wikipedia.org/wiki/End-to-end_principle
Based on my understanding of the end-to-end principle, I don't see any
significant difference in SAM vs. NAT66 WRT how much they maintain (or
violate) the end-to-end principle, as both mechanisms place some
functions/intelligence in the infrastructure.
With SAM, hosts know their global addresses, and can use them, e.g. with
SCTP or Shim6.
In my understanding, this is a difference (an an important one).
Regards,
RD
_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires
