Re: configuration: writable MIB modules versus NETCONF/YANG modules - part 2

Top posting for the hell of it.

My "problem" with these definitions are that their use of passive voice mistakes
the view from the management station and the view from the node. Maybe this is
the point Randy has. From the node's perspective, it doesn't care where the data
came from (except maybe for prioritising it over other inputs). You don't
forward packets differently on routes learnt from an IGP compared with routes
programmed by an operator.

Curiously, the definitions here seem to be different from those that I just read
on a back-channel discussion with Benoit and others where we had:

"operational data" includes both state learned from the network and the
administrative state pushed to the various routing and such processes.

"configuration data" can be persistent or ephemeral - but it's all held by the
management side, whether that is CLI or NetConf.

I am also having serious difficulty with the concept of persistence because it
can be persistent in the management station or persistent in the node. (One
hopes the latter is a subset of the former.) 

Maybe operational state can be persistent (at the node) and no-one thought to
mention it? That could certainly complete the definitions.

Anyway, is there any reason why the draft IESG statement text uses
"configuration data" in all cases except "SNMP MIB modules creating and
modifying persistent configuration"? Why "persistent" in this one case? Does it
make any difference whether SNMP is used to write persistent or non-persistent
configuration data?

And lastly...
+1 to Randy. I thought the main thrust of what we were trying to do was describe
whether SNMP Set is valuable, and if so in what context.

Cheers,
Adrian

> -----Original Message-----
> From: OPS-AREA [mailto:[email protected]] On Behalf Of Randy Bush
> Sent: 28 February 2014 16:28
> To: Benoit Claise
> Cc: ops-area
> Subject: Re: [OPS-AREA] configuration: writable MIB modules versus
> NETCONF/YANG modules - part 2
> 
> >    o  Configuration data is the set of writable data that is required to
> >       transform a system from its initial default state into its current
> >       state [RFC4741  <http://tools.ietf.org/html/rfc4741>].
> >
> >     o  Operational state data is a set of data that has been obtained by
> >        the system at runtime and influences the system's behavior similar
> >        to configuration data.  In contrast to configuration data,
> >        operational state is transient and modified by interactions with
> >        internal components or other systems via specialized protocols.
> 
> i never understood the fantasy of this dichotomy.  all (non measurement)
> data are operational.  i do not see a difference between configuring a
> bgp peer, filters, interface up/down, ...  the dichotomy is in my use
> models.  do i configure systems programatically or do i have banana
> eater keyboard jockies running my network?  the former uses netconf or
> expect and the latter use the cli.
> 
> none of them use snmp write.  i thought we made that point over a decade
> ago.
> 
> randy
> 
> _______________________________________________
> OPS-AREA mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/ops-area