Re: [ippm] John Scudder's Discuss on draft-ietf-ippm-ioam-flags-09: (with DISCUSS and COMMENT)

[Tal Mizrahi <[e-mail]>]

Dear John,

Many thanks for the comments.

We have uploaded an updated version that hopefully addresses the
DISCUSS comments.
https://datatracker.ietf.org/doc/draft-ietf-ippm-ioam-flags

Please see my responses below, marked [TM].

Please let us know if there are further comments.
Cheers,
Tal.


On Wed, Jun 29, 2022 at 8:43 PM John Scudder via Datatracker
<[e-mail]> wrote:
>
> John Scudder has entered the following ballot position for
> draft-ietf-ippm-ioam-flags-09: Discuss
>
> When responding, please keep the subject line intact and reply to all
> email addresses included in the To and CC lines. (Feel free to cut this
> introductory paragraph, however.)
>
>
> Please refer to https://www.ietf.org/about/groups/iesg/statements/handling-ballot-positions/
> for more information about how to handle DISCUSS and COMMENT positions.
>
>
> The document, along with other ballot positions, can be found here:
> https://datatracker.ietf.org/doc/draft-ietf-ippm-ioam-flags/
>
>
>
> ----------------------------------------------------------------------
> DISCUSS:
> ----------------------------------------------------------------------
>
> Thanks for this document. I have one issue I'd like to be sure we clear up.
>
> 1. In §4.1.1,
>
>    The loopback flag MUST NOT be set if it is not guaranteed that there
>    is a return path from each of the IOAM transit and IOAM decapsulating
>    nodes,
>
> This is heartwarming but I can’t see how you could guarantee this property at
> all times in any network using dynamic routing or even subject to dynamic
> conditions (and that would be all networks), and for that matter I’m not sure
> how to write code to even determine this in any general way. Is it your
> intention that this MUST NOT is directed to the operator and not to the code
> implementor? Or perhaps is it for very small values of “guarantee”? That is, is
> this an aspirational MUST and not a MUST MUST?
>
> In general it's a little problematic when we use RFC 2119 keywords in a
> protocol document, to express desires about how a protocol's operator should
> deploy it. They are at their best when used to express requirements for how a
> coder should implement the protocol. Please consider creating an operational
> considerations section, and grouping operational requirements and advice there,
> at least in that case it becomes clear to whom the RFC 2119 keywords are
> speaking.
>
> Alternately, please qualify the keywords appropriately in-line, e.g. in the
> above text you could say something like
>
>    The domain MUST be configured such that there is expected to be a return
>    path from each of the IOAM transit and IOAM decapsulating nodes; if this
>    expectation does not apply then configuration MUST NOT enable the loopback
>    flag to be set,

[TM] The point is well taken. We have adopted the text you suggested
with minor changes.

>
> To me it seems as though it might be less painful to group these into an
> operational considerations section, but whatever works for you, as long as it's
> clear.
>
> I did a cursory check over the document with this in mind, the other place I
> identified what looks like operational guidance to me is also in §4.1.1, the
> paragraph about how you "SHOULD NOT exceed 1/N of the interface capacity". At
> first blush that looks like something that could be computed automatically by
> inspection of the router's hardware, but by the time we get to the end of the
> paragraph we see that "prior knowledge about the network topology or size" is
> needed, so it must really be operational guidance. (Possibly this applies to
> the 1/N paragraphs in §4.2 and §5 also, although it's less clearly the case.)

[TM] This paragraph was rephrased to emphasize the operational aspect.


>
>
> ----------------------------------------------------------------------
> COMMENT:
> ----------------------------------------------------------------------
>
> 2. The document cites RFCs 7014 and 5475 normatively. They don't seem normative
> to me, they seem informative.

[TM] Fixed.

>
> 3. In §4.2,
>
>                                               The L-bit MUST be cleared
>    in the copy of the packet that a node sends back towards the source.
>
> This makes me wonder, does the looped back packet inherit the IP TTL/hop limit
> of the parent packet? The description of it as a “copy” makes me think it does.
> Should this be explicit?

[TM] Generally speaking, we made an effort to keep the current
document as encapsulation-independent as possible. For example, the
IPv6 Hop Limit is specific to the IPv6 encapsulation of IOAM. However,
we added the following comment:
NEW:
Creating the copy that is looped back, and specifically the
truncation, may require some encapsulation-specific updates
in the encapsulation header.


>
> NITS:
>
> 4. In §5,
>
>    This draft focuses on three possible use cases of active measurement
>
> Should be "this document focuses".

[TM] Fixed.

>
> 5. Again in §5,
>
>                                                               A selected
>       data packet that is replicated, and its (possibly truncated) copy
>       is forwarded with one or more IOAM options, while the original
>       packet is forwarded normally, without IOAM options.
>
> I think you need to delete the "that" from the first clause?
>

[TM] Fixed.

> 6. And once again in §5,
>
>    o  IOAM active measurement using replicated data packets: probe
>       packets are created by the encapsulating node by selecting some or
>       all of the en route data packets and replicating them.
>
> The 1/N requirement calls into question "or all" above, unless N=1, something
> you strongly discourage. Although you don't technically *forbid* N=1, I think
> the inclusion of "or all" creates confusion and you could and should leave it
> out while still not technically forbidding N=1.
>

[TM] The 1/N is a rate limiting recommendation is an upper bound. For
example, when the traffic rate is generally low, the "or all" is
possible without violating the 1/N recommendation. Therefore, there
does not seem to be a conflict between the two paragraphs.

> 7. In §8,
>
>                                                         The attacker can
>       potentially leverage the Loopback flag for a Distributed Denial of
>       Service (DDoS) attack, as multiple devices send looped-back copies
>       of a packet to a single source.
>
> The use of "source" is odd here. By the nature of an attack, the looped-back
> copies wouldn't be targeted at the actual source of the packets. Possibly
> "target" or even "victim"?

[TM] Fixed.

>
>
>