[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: psamp vocabulary
Rae McLellan wrote:
> >>Except for one case, I don't believe applying the selectors in any
> >>particular order produces different results. Though early out for
> >>performance might be something a compiler could achieve, the end
> >>result of sampled traffic remains the same. I don't really care
> >>what syntax is used to specify the selectors or how they are combined.
> >>But the functionality of union and conjunctions is important.
> >>The only place where the order of the selector rules matters is
> >>when performing the "sample K packets every N" type selector.
> >>This is because there is a difference in the range over which
> >>the N is sampled. If this type of sampler is applied first, N
> >>ranges over the entire input stream. But if it is applied last,
> >>then N only applies to those packets that have managed to pass
> >>through any previous selector functions. Each will produce a
> >>different sample stream.
> > I think it is not true that in most cases the sequence in which
> > selectors are applied does not matter. If you perform any random
> > sampling and then run any deterministic function on the subset you will
> > always have different packets in the sample than you would get if you
> > applied the selectors in the reverse sequence.
> oops, you're entirely right. Its not the just the "sample K packets
> every N" type selector, but also the random ones in the third group
> that depend on order. Its my fault, I've been deprecating random
> sampling in favor of deterministic hash functions so much, that I'd
> forgotten about that problem. If psamp can eliminate this group
> entirely, there would be no issue with selector ordering. (i think)
> correct me, if I'm wrong.
I wonder whether this "invariance" in the selected packet stream when swappig
order of selectors can be considered a generally "desirable" property. While
it is for some
application (e.g trajectory sampling), for some other (e.g. traffic
profiling) what matters is only that, independently of selector ordereing,
1) the amount of selected packets doesn't (significantly) change
2) the selected sample has a distribution of the metric of interest similar
to that of the parent population, so that it can be correctly inferred
analyzing the sample.
> Rae McLellan
> to unsubscribe send a message to email@example.com with
> the word 'unsubscribe' in a single line as the message text body.
> archive: <http://ops.ietf.org/lists/psamp/>
to unsubscribe send a message to firstname.lastname@example.org with
the word 'unsubscribe' in a single line as the message text body.