Re: [Ace] I-D Action: draft-ietf-ace-cmpv2-coap-transport-01.txt

[Mohit Sahni <[e-mail]>]

Hi David
Many thanks for the comments, please see my response inline, I will
wait for couple of more days for any new comments and will publish a
new version of draft with the name change (cmpv2 -> cmp) and changes
based on your feedback.

Regards,
-Mohit


On Thu, Apr 29, 2021 at 6:34 AM David von Oheimb
<[e-mail]> wrote:
>
> Hi Mohit,
>
> a couple of comments for the WGLC also from my side.
>
> Abstract:
>
> IMO it would be nice (but it's of course not strictly needed) to refer to the Lightweight CMP profile already in the abstract,
> maybe this way after the first sentence:
>
>      It details the CoAP transfer option mentioned in the Lightweight CMP Profile.
>
[M.S.]: Agreed.
> Section 1:
>
> encryption of messages -> protection of messages    (because authenticity is the predominant requirement)
>
[M.S.] Agreed
> between CAs -> between RAs                          (between CAs makes little sense, but there may be more than one RA involved)
>
[M.S.] Cross certificate signing requires two CA's to communicate with
each other (https://tools.ietf.org/html/rfc4210#section-6.6).
> Section 2.6:
>
> the Block-Wise transfer [RFC7959] mode
>    MUST be used for the CMP Transactions over CoAP
>
> I do not have a strong opinion here, but I fear that strictly requiring block-wise transfer may needlessly exclude simple implementations,
> which may be sufficient in scenarios where the payloads are known to be rather small.
> Writing SHOULD or RECOMMENDED would state that implementors can deviate from the recommendation,
> but only if they are aware of the consequences and are willing to cope with them.
> So what you could do - if others agree - is to replace
>
>    In order to avoid IP fragmentation of messages exchanged
>    between EEs and RAs or CAs, the Block-Wise transfer [RFC7959] mode
>    MUST be used for the CMP Transactions over CoAP.
>
> by a strengthened recommendation with a motivation/warning, e.g.,
>
>    Block-wise transfer [RFC7959] mode
>    SHOULD be used for the CMP Transactions over CoAP.
>    This is strongly recommended to avoid IP fragmentation of messages
>    and the block-wise option is a critical option as per RFC 7959.
>
[M.S.] In my opinion if we make this optional, then we will end up
with implementations that will have to support both block-wise and non
block-wise mode and may cause several interoperability issues.
> Section 3:
>
> Nice to see that you streamlined the text regarding DTLS.
>
> Section 4:
>
> cross protocol proxy -> cross-protocol proxy
[M.S.]  Agreed.
>
> pre configured servers -> pre-configured servers
>
[M.S.]  Agreed.
> Section 5:
>
>  In order to protect themselves against DDoS attacks, the
>    implementations SHOULD avoid sending or receiving very small packets
>    containing partial CMP PKIMessage data.
>
> Sounds good, but the point is not distributed DoS (only) but DoS in general, so: DDoS -> DoS
> and there is no real protection against DoS, just reduction of risks they impose.
> Moreover, the recipient has little influence on the size of packets.
> I'd further suggest streamlining the sentence, arriving at, e.g.,:
>
[M.S.] DDos->Dos agreed

> "In order to reduce the risks imposed by DoS attacks,
> implementations SHOULD minimize fragmentation of messages,
> i.e., avoid packets containing partial CMP PKIMessage data."
>
[M.S.] Agreed.
> And better starting a new paragraph thereafter because using a CoAP-to-HTTP proxy is a different topic.
[M.S.] Agreed.
>
> Regards,
>
>     David