Re: [xmpp] Fwd: I-D Action: draft-saintandre-xmpp-tls-00.txt

[Peter Saint-Andre <[e-mail]>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 9/9/13 10:51 AM, Dave Cridland wrote:
> On Mon, Sep 9, 2013 at 5:35 PM, Peter Saint-Andre
> <[e-mail] <mailto:[e-mail]>> wrote:
> 
> There's also a htmlized version available at: 
> http://tools.ietf.org/html/draft-saintandre-xmpp-tls-00
> 
> 
> Thanks for writing this, it looks like an excellent start.

You inspired it because as I recall you suggested that we pull some of
the security recommendations out of draft-ietf-xmpp-3920bis and into a
separate spec that we could update more readily. :-)

> o  Add information about the rationale for each recommendation, 
> perhaps in an appendix.
> 
> That seems reasonable, but I don't think there's a need to describe
> each recommendation individually.

Agreed, and the justification for some of them is already available
elsewhere. It seems that draft-sheffer-tls-bcp might evolve in that
direction, too.

> o  Recommend a specific ciphersuite or a small number of 
> ciphersuites?
> 
> Picking a small selection of SHOULD or MUST implement seems fair.
> 
> o  Provide recommendations regarding key lengths?
> 
> I think you already do for ephemeral keys, but public key lengths
> yes.

I'm thinking of the discussion here:

http://www.ietf.org/mail-archive/web/perpass/current/msg00062.html

> o  Discuss TLS compression vs. application-layer compression?
> 
> It's not obvious it's a security issue, but yes.

Well, there is the CRIME attack against some technologies that use TLS
compression, however it's not clear to me if that applies to XMPP.

> The issue remains that compressing streams takes a serious wodge of
> data, but it's desirable.
> 
> Also, I think we should discuss the use of anonymous (or
> unauthenticated in general) TLS; I've been convinced recently that
> using even a non-channel-binding authentication and unauthenticated
> TLS is substantially better than nothing. It's also what the
> deployed base actually does.

Could you explain a bit more what you mean by unauthenticated TLS?

> Finally, I'd be keen to see this adopted by the working group if
> that's on the table.

I leave that up to the chairs. :-)  Depending on how you read the
charter, it might or might not be in scope.

Peter

- -- 
Peter Saint-Andre
https://stpeter.im/


-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.19 (Darwin)
Comment: GPGTools - http://gpgtools.org
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQIcBAEBAgAGBQJSLgB1AAoJEOoGpJErxa2pL5IP/j6S7Akfij35ZoZnm8GQrw+t
J/sEKsZjRmICAu0hXUfhTPl2cZKcYF874M8WlypnzD4noA+UzHwvKWBjN2m/1Lwa
xCHeVWmwO88fG6xQC5cZxycUlyvdDb2foho672OsFXk58qIOVoBSs74YPANQbfKB
7396vlwHtqnygI1FEqyQYjdBtZztlceRHJ7G0DUGxpjoEG2UMGIQVLVMrqLUIeVE
YLIFVYtd6P+uKAIqwFOt4LjbCbozAX5jSr2W1lUUz7bo3sOrB1TpWeuyOpvHZulm
Kp7qhxsZDrQWeupbbtXP4rwnOXipVfUCY7mBpdRl0FIdKV5xqrdyBzNGwvY8sVxX
OmVeF1+53Sm2hftUoxbDWoFcoI3CoqzeXSS+ayz6L+FW+Jj3aEmzp/caJNNsjHM9
3In5IxOl4IwHRKL/IhRWZRN9vfETBa5ncb4EbLvhr0tzFCnXK2E5nb3bpUBTVTB6
Jlqv8dmiAeJFGESz3R1RYr6uelpC+WgTGMXP1Oz8uSlZxNnhSIeaqqLXkzYz0hmN
ZjpL2X+J9kldwUflBF5ARkakvbo1LCPxFZdoEd+dixiYO7YN6OAn4ZK6vzTvJiOa
RblKSM8HVBcBID1B3cnQiStfqUYsEGtonKwV7LKJ0Pem/tzoT7jgZkAhiLDR0wzf
Q/ASo78kt3ZeamgXjqHe
=wM8x
-----END PGP SIGNATURE-----