Hi Hilarie,
Thanks for sticking with this.
> Minor grammar issue re the terms Network Telemetry, Network Analytics
> and Network Observability. These are "processes", and I think
> "Network Analysis" and "Network Observation" are better terms for
> processes. For telemetry, the definition seems ambiguous as to
> whether is is a process or the data collected by the process. If it
> is a process, then the title should be "Network data collection" or
> something similar.
I'm pretty sure the word "telemetry" denotes a process.
To quote Wikipedia (so it must be true), "Telemetry is the in situ collection of measurements or other data at remote points and their automatic transmission to receiving equipment for monitoring."
The text is, I think, unambiguously describing a process.
Propose no change.
> "System: An assembly of components that exhibits some behavior." How
> about simply "a collection of components in a network"? It's more
> specific and introduces no undefined terms.
There are three points here:
- s/assembly/collection/
I can't get overly excited about this.
Collection has a less ordered feel.
Assembly is more commonly applied to people.
Assemblage would be more correct than assembly, but simply means collection.
I'm going to make this change unless the WG shouts.
See also in the definition of Resource.
- d/that exhibits some behavior/
This is probably redundant, but we were attempting to express what it is that is
interesting about the collection of components.
I suppose that, pedantically, "no behavior, is still a behavior". That truly makes
the text redundant.
I'm going to delete this, unless the WG shouts.
- a/in a network/
Conversely, I think this would be a redundant addition. It should be clear
that this work (all IETF work?) is limited to networking.
Additionally, the term "network" actually opens up all sorts of ambiguity.
I am not going to make this change.
> "Value: ... on a continuous variable (e.g., an analog measurement)"
> The grammar is wrong, and the "continuous variable" terminology is,
> as before, just wrong.
I find no fault in the grammar.
However, s/on/of/ might make it easier.
The discussion of continuous variables (analog measures) in a digital system may be verging on the philosophical. Zeno would, I feel, have delighted in our struggles.
Consider a reading of temperature. From an analog device, we will get a continuous variable, but (clearly?) in order that it can be processed by a digital system it will be converted into some set of discrete values: there will be an end to the number of decimal places. There is, however, I believe benefit in listing the different classes of input to the generation of a Value.
Not planning a change.
> "Not all Changes are noteworthy (i.e., they do not have Relevance)."
> That comment should be part of the section defining "Relevance".
OK added it there, too.
> "Event: The variation in Value of a Characteristic of a Resource at a
> distinct moment in time (i.e., the period is negligible)." What
> is the point of this? Any variation is a measured value, and a
> measurement takes place at "a distinct moment in time."
I think you captured the point by including the word "measured".
Hmmm. Consider the values on and off at times t1, t2, t3. {t1, on}, {t2, off}, {t3, on}
With a sampling rate of > t3-t1 no change is recorded.
However, there are events we want to notice (in particular, the transition to off).
The transition may generate an event from some system regardless of the over-arching sampling process.
I think the challenge lies in looking at the (fractal!) nature of the monitoring systems.
Compare with a human who may look at the clock from time to time, and the alarm in the clock which may sound even when the human is not looking at the clock.
> "While a State may be observed at a specific moment in time, it
> is actually determined by summarizing measurement over time in
> a process sometimes called State compression." I don't think
> that's the normal meaning of "state compression".
Do you have a reference for the normal meaning?
If the mention of state compression is adding confusion rather than clarity, we would surely remove it.
> This version of the document is much improved over -07, and it doesn't
> seem to have any negative security implications.
Thank you for your help in achieving that.
Best,
Adrian
--
last-call mailing list -- last-call@xxxxxxxx
To unsubscribe send an email to last-call-leave@xxxxxxxx
