RE: Methods in the NIM requirements

["Lakshmi Raman" <lraman@telcordia.com>]

Comments in line.
Lakshmi




"Andrea Westerinen" <andreawest@mindspring.com> on 05/01/2000 04:21:25 AM

To:   "Joel M. Halpern" <joel@mcquillan.com>, nim@ops.ietf.org
cc:    (bcc: Lakshmi Raman/Telcordia)
Subject:  RE: Methods in the NIM requirements




Comments inline.
Andrea

> -----Original Message-----
> From: owner-nim@ops.ietf.org [mailto:owner-nim@ops.ietf.org]On Behalf Of
> Joel M. Halpern
> Sent: Monday, May 01, 2000 12:18 AM
> To: nim@ops.ietf.org
> Subject: RE: Methods in the NIM requirements
>
>
> I would consider an interesting separate question what real-time
> information the information model needs to represent.  We are not
> after all trying to take over the SNMP / RTFM job of deciding what
> statistics need to be collected.

<arw> I would not want to second guess what statistics are needed.  But, in
as much as an info model should reflect pertinent data in a
repository/protocol independent manner, you may include SNMP statistics and
other real-time data in the model. This makes independent reference to the
data possible, for use (for example) in policies.

<LR> I would consider also including in the model what statistics are
to be collected. In the Telecom world, we collect a lot performance data such as
severly error second, loss of pointer (SDH/SONET) which we do model.
In addition we also model threshold values and so if the values
being collected exceed threshold, you send a quality of service alarm.
These parameters are being collected at different intervals and real time.
However, we do have to model them so that one can define what
information is to be collected.

> Once we have the requirements and language selected, an interesting
> question is whether we want to separately represent "should be" and "is"
> states about devices.  My first reaction is that we need to represent
> "should be" and that other systems already deal with "is".  Sorry for the
> digression, lets try dealing with the ducks in order.

<arw> I would agree that this is an interesting question.  But, my first
reaction would be different than yours.

<LR> I would also say my reaction will be different. I can think of
cases where both information should be modeled.
> Joel
>
> At 06:05 PM 4/29/00 -0700, Andrea Westerinen wrote:
> >... nor any real-time/dynamic DATA
>
>
>