[anonsec] feedback on draft-williams-btns-00.txt

[Joe Touch]

Hi, Nico (et al.),

Here's some comments on the draft. Overall, mostly editorial or
clarification.

Joe

------

sec 1:

It seems like "MITM" might be more generally "on-path" attacks, of which
MITM is a subset (where the attacker can remove messages, as well as
read and inject them)

If 2401bis is strictly required (vs. 2401), it might be useful to adjust
the title (not sure, though). I'd prefer if there were a way to refer to
2401 processing, even if in an appendix, though I don't know if that's
feasible.

'opportunistic' might be replaced with 'fallback', since 'opportunistic
IPsec' has another meaning (proactive keying).

FWIW, fallback isn't mentioned in the P&AS - should it be?

sec 2:

is 'coerce' common terminology? I didn't see it in the IKE or IKEv2
docs; I'm not sure what process is implied. it might be useful to
explain this to be more specific.

why must the cert payloads be generated for the purpose of being used
this way? unless I'm misreading it, it seems like this means we can't
have a server that some endpoints can validate and others cannot, which
uses a single key (I don't know why that would matter - if it does, it'd
be useful to explain, and if not, it'd be useful to avoid the restriction)

s/assymentric/asymmetric/

---


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 254 bytes
Desc: OpenPGP digital signature
Url : http://www.postel.org/pipermail/anonsec/attachments/20051005/f7b825ef/signature.bin

[Sam Hartman]

Joe> If 2401bis is strictly required (vs. 2401), it might be
Joe> useful to adjust the title (not sure, though). I'd prefer if
Joe> there were a way to refer to 2401 processing, even if in an
Joe> appendix, though I don't know if that's feasible.



Hi. I've been working with Nico on trying to understand the
implications of his draft for 2401bis. I've also been following the
Kink working group's efforts to work with both 2401bis and 2401.

My reasonably strong belief is that we have to do more than twice the
work if we want to support both 2401bis and 2401. we need to do the
analysis separately for each model (they are that different) and we
then need to make sure the results are consistent.

We must support 2401bis. I argue that we should not spend effort on
2401.

I understand David's concerns WRT iscsi and IKE V1. I don't think
only thinking about the 2401bis architecture precludes IKE V1 from
being considered.

--Sam

[Nicolas Williams]

Post by Sam Hartman
Hi. I've been working with Nico on trying to understand the
implications of his draft for 2401bis. I've also been following the
Kink working group's efforts to work with both 2401bis and 2401.

BTW, I'll write up our discussion and send notes to the list tomorrow.

Nico
--

[Joe Touch]

Post by Sam Hartman
Joe> If 2401bis is strictly required (vs. 2401), it might be
Joe> useful to adjust the title (not sure, though). I'd prefer if
Joe> there were a way to refer to 2401 processing, even if in an
Joe> appendix, though I don't know if that's feasible.
Hi. I've been working with Nico on trying to understand the
implications of his draft for 2401bis. I've also been following the
Kink working group's efforts to work with both 2401bis and 2401.
My reasonably strong belief is that we have to do more than twice the
work if we want to support both 2401bis and 2401. we need to do the
analysis separately for each model (they are that different) and we
then need to make sure the results are consistent.
We must support 2401bis. I argue that we should not spend effort on
2401.

BTNS affects the SPD in 2401/2401bis in similar ways (and not much
else); it's not clear that there will be double the effort required to
address both (as is the case with KINK). It'd be useful to understand
that before abandoning 2401.

Post by Sam Hartman
I understand David's concerns WRT iscsi and IKE V1. I don't think
only thinking about the 2401bis architecture precludes IKE V1 from
being considered.

It seems like 2401bis refers only to IKEv2 (end sec 3.2 of 2401bis).

Joe