[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Feedback -1



Christian,

thanks for your comments. Responses inline.

Nick

Cristian Estan wrote:
[..]
> As probably all drafts in this stage, this one is often too
> general. For example the last paragraph of example 1 in section 1
> could definitely be made more concrete.
> 
> Often the draft refers to "sampling parameters" and "sampling data"
> where in fact more general phrase should be used given that the draft
> permits hashing and filters also. For example the first two of the
> last three bullets in the goals chapter.

We need to employ better language to distinguish between statistical 
sampling in particular, and selecting packets by whatever method 
(e.g. statistical sampling, filtering, hash-based) in general. 
"Sampling parameters" and "sampling data" were meant in the second, 
more general, sense.

> 
> The last paragraph of section 3.2 raises some issues worth
> discussing. Does it mean that measurement cannot rely on things such
> as destination prefix (or AS) which are probably available in the
> router interface card? If so this point should be argued more
> thoroughly. The "Deriving Traffic Demands for Operational IP Networks:
> Methodology and Experience" has a very nice argument for separating
> the routing infrastructure from the measurement part. Even though I
> agree with that argument, applied strictly it would prohibit things
> such as NetFlow aggregation. Do we want to do this?
> 

The point is not to require that packet selection be dependent on
routing state, since this places an architectural constraint on the
router. 

The draft isn't concerned with what on-board operations (such as 
aggregation) you might do with packets after they have been selected,
although how the selected packets are presented to such applications
is in scope. However, the draft wouldn't preclude aggregation at the
router: this would be done after the packets have been selected.

--
to unsubscribe send a message to psamp-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/psamp/>