Re: [secdir] sector review of draft-ietf-pcp-server-selection-07

["Prashanth Patil (praspati)" <[e-mail]>]

Hi Med,
Given that this is a recommendation for efficiency, we could perhaps
change 'SHOULD' to a 'MAY'.

We only introduced this because it's easier on the client to use the same
mapping nonce when it tries to connect with each address of the same
server in a sequential manner.



-Prashanth

On 1/5/15 1:04 PM, "[e-mail]"
<[e-mail]> wrote:

>Dear Chris,
>
>Thank you for the review.
>
>Nonce validation checks are used when operating in the simple threat
>model as discussed in RFC6887. Adding a reference to section 18.1 of
>RFC6887 would be sufficient IMHO. PCP authentication will be needed if
>operating in the Advanced Threat Model.
>
>OLD:
>For efficiency, the PCP client SHOULD use the same Mapping Nonce for
>requests sent to all IP addresses belonging to the same PCP server.
>
>NEW:
> For efficiency, the PCP client SHOULD use the same Mapping Nonce for
>requests sent to all IP addresses belonging to the same PCP server. As a
>reminder, nonce validation checks are performed when operating in the
>Simple Threat Model (Section 18.1 of [RFC6887]) to defend against some
>off-path attacks. 
>
>Better?
>
>I already fixed the nits in my local copy.
>
>Thank you.
>
>Cheers,
>Med
>
>-----Message d'origine-----
>De : Chris Inacio [mailto:[e-mail]]
>Envoyé : dimanche 4 janvier 2015 07:57
>À : [e-mail]; [e-mail];
>[e-mail]
>Objet : sector review of draft-ietf-pcp-server-selection-07
>
>
>
>I have reviewed this document as part of the security directorate¹s
>ongoing effort to review all IETF documents being processed by the IESG.
>These comments were written with the intent of improving security
>requirements and considerations in IETF drafts.  Comments not addressed
>in last call may be included in AD reviews during the IESG review.
>Document editors and WG chairs should treat these comments just like any
>other last call comments.
>
>Generally the document is in good shape, and I would like to see one
>minor issue at least commented upon.
>
>I have a single security related comment on this draft; the last sentence
>of section 3:
>
>> For efficiency, the PCP client SHOULD use the same Mapping Nonce for
>>   requests sent to all IP addresses belonging to the same PCP server.
>
>Normally, I would simply say this is a crazy recommendation.  But after
>looking a little into what the Nonce is used for in the PCP protocol, I
>am slightly less distraught.  This Nonce does not appear to necessarily
>provide any huge amount of security except allowing the client to
>generate a unique token per PCP server.  Presumably there is a general
>MITM attack on the PCP protocol related to the Nonce as a transaction ID
>which is prevented by using other security protocols, TLS, etc.  (And
>another well known attack with the THIRD_PARTY option and lack of
>authenticationŠ) Therefore, this Nonce is critical as a synchronization
>point between the client and the potential PCP server.  It would be nice
>(assuming all that is correct) to make that clear in the document,
>especially with a recommendation to reuse the Nonce.
>
>
>Nits:
>
>In Figure 1, the lines are not aligned to the ³+² on the diagrams.
>
>In Figure 3, ³rtr1² is missing a ³+² on the right side connection from
>the top.
>
>
>--
>Chris Inacio
>[e-mail]
>
>
>