Re: [Softwires] I-D Action: draft-ietf-softwire-lw4over6-06.txt

[Wojciech Dec <[e-mail]>]

Ted,
May I kindly ask that you read my comments before starting apparently
rhetorical discussions of the type that I didn't intend, and that do not
progress things?

In summary: I said that having a lw46 draft/solution is fine, but it is
clear that there is significant technical overlap between the two (use of
PSID w/ MAP algorithm, IP address embedding, NAT44 port sharing). At this
stage it doesn't matter how it came to be that we have two solutions+
(personally I believe that it was a case of WG leadership gone totally
wrong at the time).

What matters is that it is readily apparent that any divergences in the
area of IP tunneling and ICMP handling are not warranted - or if they are,
then the draft doesn't clarifyr why that is so. What also matters is to
make this clear to implementers. Yes, I do belive that people outside of
this WG will be interested to implement devices that cover both solutions
and minimize the work in doing so, which doesn't take away from them the
ability to do just one.

As a separate set of feedback points, a number of requirements in the lw46
draft do not appear to belong there (e.g they restate, selectively, NAT44
best practices). The WAN IP address selection appears either a) broken (e.g
author's didn't answer how to configure the IPv6 address in case of
multiple prefixes, or if the WAN interface doesn't have a global IPv6
prefix) or b) limited to a specific deployment. This needs to be clarified.


On 25 February 2014 19:29, Ted Lemon <[e-mail]> wrote:

> On Feb 25, 2014, at 1:21 PM, Tom Taylor <[e-mail]> wrote:
> > I know this is a political rather than a technical point, but one could
> sort of reverse the claim and suggest that if 1:1 mode is desired, LW4o6 is
> the better way to go.
>
> Indeed, if all one needs is 1:1, one might prefer lw4over6 since it only
> has the one mode.
>
>