From: Adrian Farrel <adr...@olddog.co.uk>
Sent: 25 October 2024 12:59
Thanks Tom,
This is useful.
<tp>
Top posting what would otherwise require you to scroll all the way to the
bottom.
My last comment
> or, choosing one of many,
>
> Path: OAM in relation to a path
> Packet: OAM in relation to a user data packet.
really is one of many - that format where a key top-level concept has not been
explicitly specified rendering the lower level term ambiguous or contradictory
does occur in several places; now you have seen it for yourself in one place
you might see it in others.
Tom Petch
> e.g.
>
> Active-Passive OAM.
> Active-Hybrid OAM.
> Hybrid-Passive OAM.
> Active-Hybrid-Passive OAM.
>
> I see no definition of these categories nor any suggestion as to what I
should do with
> these phrases. Should authors start classifying their OAM based on their
understanding
> of these words without further explanation? Or what?
OK. So we currently have...
Compound OAM:
Uses a combination of at least two of Active OAM, Passive OAM, and
Hybrid OAM (i.e., a combination of atomic OAM packets, data packet
modification for OAM, and no explicit OAM). Note that [RFC7799]
also uses the term "Hybrid" to refer to metric types in-between
active and passive, for OAM there are no in-betweens per se, only
active, passive, hybrid, or a compound combination.
Compound OAM can be characterized in a more explicit way, for
nuanced use-cases:
* Active-Passive OAM.
* Active-Hybrid OAM.
* Hybrid-Passive OAM.
* Active-Hybrid-Passive OAM.
...and there are (I think) clear definitions of each of Active, Passive, and
Hybrid OAM.
So what appears to be missing is a little more description of what it means
to combine two or more OAM methods.
Perhaps:
ADD
When two or more OAM types are combined, it is simply the case
that both OAM types are used at once. For example, in Active-Passive
OAM both specific OAM packets and the observation of data packets
may be used to provide information about the data flow in the network.
Each component may supplement the total data.
END
We can also make clear that the naming is according to the combinations.
Thus:
OLD
Compound OAM can be characterized in a more explicit way, for
nuanced use-cases:
NEW
Compound OAM can be characterized in a more explicit way, for
nuanced use-cases, and named according to the OAM types that
are combined:
END
> or, choosing one of many,
>
> Path: OAM in relation to a path
> Packet: OAM in relation to a user data packet.
>
> What are you trying to convey here? I have an understanding of
> what a packet is and what a path is but what this might be saying
> about a path or a packet defeats me.
Funny. These just bit me when I was re-reading. The terms are too terse. We
are (clearly) not trying to redefine the meaning of the word "path" in this
document!
I think we should have:
Path OAM: OAM in relation to a path.
Packet OAM: OAM in relation to a user data packet.
Packet Treatment OAM: OAM in relation to the treatment of user
data packets
Combined OAM: OAM in relation to multiple criteria.
> HTH
It does.
Adrian
_______________________________________________
OPSAWG mailing list -- opsawg@ietf.org
To unsubscribe send an email to opsawg-le...@ietf.org
