Re: [IPFIX] Fwd: [RFC 3917]Overload & Reliability requirements of Metering process

[sujay gupta <sujay.ietf at gmail.com>]

Hi Paul,

Thanks,

IMO it is safer to call the period as an outage period.
But then if the reset period is short, and the collector has
sufficient data points it may
very well model and interpolate the missing data. And if it does this
it may as well choose
to ignore the partial flows residing during the reset period, even if
the switch attempts to send it.

But then your email is helpful to point out collectors may not resort
to stats modeling.

BR,
-Sujay

On Wed, Oct 21, 2009 at 5:22 PM, Paul Aitken <paitken at cisco.com> wrote:
> sujay,
>
>> So what I am proposing ( and is possible) is that under these
>> conditions like O/L , soft restart the system (switch/router) can
>> easily send to the collector  at least the "duration under which it
>> was in the O/L/reset condition" . This information I am thinking would
>> be useful as a rough estimate of monitoring lost, canbe converted in
>> to number of packets not accounted for
>
> No, it can't.
>
> There are too many variables which the collecting process would have to
> know. eg how many active monitoring processes are there on the exporting
> device, and how much traffic are they each measuring on average?
>
> The delay between measuring and exporting, and the flow lifetime on the
> switch will also be relevant, since you'll export some partially filled
> flows for some period of time after the reset.
>
>
>> or may be used as an unaccounted period by the collector.
>
> I think that's the only possibility. Let's not try to guess how much traffic
> was lost.
>
> Cheers,
> P.
>
> --
> Paul Aitken
> Cisco Systems Ltd, Edinburgh, Scotland.
>