Re: [Ntp] Tsvart last call review of draft-ietf-ntp-packet-timestamps-08

[Tal Mizrahi <[e-mail]>]

Hi Ian,

Thanks for the thorough review.

Please see some comments inline.

On Mon, Mar 2, 2020 at 6:47 PM Ian Swett via Datatracker
<[e-mail]> wrote:
>
> Reviewer: Ian Swett
> Review result: Ready with Nits
>
> Overall, this draft looks very good.  It uses 'should' non-normatively more
> often than I'd like.  In some cases, I wonder if they should be normative
> SHOULDs.

The draft only uses two instances of (uppercase) normative requirement
language words, one in Section 3 (which defines the timestamp
template) and the other in Section 9 (Security Considerations). This
is intentional, since these are exactly the requirements that the
reader is expected to comply to when defining a timestamp.

>
> Section 1.2(Scope)
>
> Timestamps are also used in RTP(RFC3550).  Based on your document, I think
> these are out of scope, because they're 'application' timestamps, not network
> timestamps?  I was unclear on this until I read the draft, so if they are out
> of scope, it would have helped me if you said that application timestamps, such
> as RTP are out of scope.

Right, timestamps are used in RFC 3550, and actually this is already
mentioned in Section 6 of the draft (RFC 3550 has actually two type of
timestamps, the "NTP" timestamp, and the "RTP" timestamp).

>
> Section 3, under "Epoch:"
>
> Since on power up is an example of a relative timestamp and not the only
> possible relative timestamp, I'd suggest ", in which the epoch could be the
> time at which..."  The current reading implies that all relative timestamps are
> due to the start being the power up time, and I don't think that's true.
>

Agreed.

> Section 4
>
> I'd change "Specifically, if the network..." to "For example, if the network..."

Agreed.

>
> Section 9, Security Considerations
>
> The sentence beginning with "Timestamps can be spoofed or modified..." implies
> that all network timestamps have this property, but you rightly point out later
> in the paragraph that one can add a MAC to prevent modification.  I'd suggest
> something like "In some cases, timestamps can be..." to avoid the implication
> that all of them can be spoofed or modified.
>

Agreed.


Cheers,
Tal.