Hi Yoshi,
Thank you for the review.
Please see inline.
Cheers,
Med
> -----Message d'origine-----
> De : Yoshifumi Nishida via Datatracker [mailto:noreply@xxxxxxxx]
> Envoyé : dimanche 31 mars 2019 10:53
> À : tsv-art@xxxxxxxx
> Cc :
draft-ietf-dots-signal-channel.all@xxxxxxxx;
ietf@xxxxxxxx;
> dots@xxxxxxxx
> Objet : Tsvart last call review of draft-ietf-dots-signal-channel-31
>
> Reviewer: Yoshifumi Nishida
> Review result: Almost Ready
>
> This document has been reviewed as part of the transport area review team's
> ongoing effort to review key IETF documents. These comments were written
> primarily for the transport area directors, but are copied to the document's
> authors and WG to allow them to address any issues raised and also to the
> IETF
> discussion list for information.
>
> When done at the time of IETF Last Call, the authors should consider this
> review as part of the last-call comments they receive. Please always CC
> tsv-art@xxxxxxxx if you reply to or forward this review.
>
> Summary: This document is almost ready for publication, but it will be better
> to clarify the following points.
>
> 1: "it is out of scope of this document to specify the behavior to be
> followed by a DOTS client to send DOTS requests when multiple
> DOTS servers are provisioned."
>
> I'm not sure why it is out of scope. Does it bring a certain
> complexities
> to the protocol?
> Or, does it simply mean it is up to implementations?
[Med] This is because of specific considerations such as those discussed in draft-ietf-dots-multihoming. We don't want to overload the base spec with such considerations (discussed in separate documents). The same approach is followed for server discovery (draft-ietf-dots-server-discovery).
Got it. I think it would be useful for readers to mention it in the draft.
[Med] I updated the text as follows :
Likewise, it is out of scope of this document to specify the behavior
to be followed by a DOTS client to send DOTS requests when multiple
DOTS servers are provisioned (e.g., contact all DOTS servers, select
one DOTS server among the list). Such behavior is specified in other
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
documents (e.g., [I-D.ietf-dots-multihoming]).
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> 2: "The DOTS client periodically repeats the mechanism to discover whether
> DOTS signal
> channel messages with DTLS over UDP becomes available from the DOTS
> server.."
>
> -> Does this mean DOTS clients will not repeat this when it already
> has
> DTLS over UDP connection?
> What about if the client has DTLS over UDPv4? Does it try to check
> DTLS over UDPv6? Also, is this logic MAY or SHOULD or don't want
> to
> specify?
[Med] The client will retry when the cache flushes out:
Note that the DOTS client after successfully establishing a
connection MUST cache information regarding the outcome of each
connection attempt for a specific time period, and it uses that
information to avoid thrashing the network with subsequent attempts.
OK, but if you retry anyway after a certain time, why do you need to cache the results?
[Med] “to avoid thrashing the network with subsequent attempts” as the text says.
Also, caching avoids the inconvenience of selection during attack times. Typically, the client executes the HE procedure when it initializes and uses the outcome of the procedure for subsequent DOTS signal exchanges during a certain period. Please refer to
this text in the draft:
To overcome these connection setup problems, the DOTS client attempts
to connect to its DOTS server(s) using both IPv6 and IPv4, and tries
both DTLS over UDP and TLS over TCP in a manner similar to the Happy
Eyeballs mechanism [RFC8305]. These connection attempts are
performed by the DOTS client when it initializes, or in general when
it has to select an address family and transport to contact its DOTS
server. The results of the Happy Eyeballs procedure are used by the
DOTS client for sending its subsequent messages to the DOTS server.
> 3: "DOTS agents SHOULD follow the data transmission guidelines discussed
> in Section 3.1.3 of [RFC8085] and control transmission behavior by
> not sending more than one UDP datagram per round-trip time (RTT) to
> the peer DOTS agent on average."
>
> -> How about TCP connections?
[Med] TCP does not have the issue with non-confirmable messages. FWIW, RFC8323 states the following:
The request/response interaction model of CoAP over TCP is the same
as CoAP over UDP. The primary differences are in the message layer.
The message layer of CoAP over UDP supports optional reliability by
defining four types of messages: Confirmable, Non-confirmable,
Acknowledgment, and Reset.
...
Since TCP eliminates the need for the message layer to support
reliability, CoAP over reliable transports does not support
Confirmable or Non-confirmable message types.
I see. Thanks for the clarification.
