Re: [netmod] draft-ietf-netmod-entity issue #12

[Martin Bjorklund]

Juergen Schoenwaelder  wrote:
> On Thu, Dec 15, 2016 at 04:03:40PM +0100, Martin Bjorklund wrote:
> > Hi,
> > 
> > Issue https://github.com/netmod-wg/entity/issues/12
> > 
> >   entConfigChange is useful but maybe covered by netconf-config-change
> >   notification, or pub/sub. it is also a bad name, since it fires when
> >   the opstate changes...  Need to think about which notifs to define.
> > 
> > Currently we have three notifications:
> > 
> >   harwdare-state-change
> >   hardware-state-oper-enabled
> >   hardware-state-oper-disabled
> 
> It seems harwdare-state-change is not netconf-config-change. The
> harwdare-state-change indicates that the '/hardware-state/component'
> list has changed, i.e., operational state has changed not config.  (It
> is always important to read the descriptions and not to draw wrong
> conclusions from the fact the SNMP notification was called
> entConfigChange.)

Yes.

> > Jason Sterne also made the observation on the ML that maybe pub/sub is
> > a better mechanism for these simple notifs.
> 
> Not sure what 'better' means. I assume that generating these specific
> notifications can likely be done way more efficiently than doing the
> same via a pub/sub interface unless the pub/sub backend code is really
> really smart.

Yes, I share this concern.  I don't think it would be a good idea to
remove all notifications and just provide state nodes, and then assume
that clients write clever filters in order to get the notificaitons
they want.

However, in this particular case, the notifications
hardware-state-oper-enabled and hardware-state-oper-disabled are
simple in the sense that they defined to be generated when a certain
node changes its value.

> That said, I would not be surprised if application
> writers will use pub/sub for everything because the hammer makes
> everything look like a nail.

We'll see.  Specific notifications are often designed to deliver
related information, and this might not be to easy to do with a
generic solution like pub/sub.


/martin

___
netmod mailing list
netmod@ietf.org
https://www.ietf.org/mailman/listinfo/netmod

Re: [netmod] draft-ietf-netmod-entity issue #12

[Juergen Schoenwaelder]

On Thu, Dec 15, 2016 at 04:03:40PM +0100, Martin Bjorklund wrote:
> Hi,
> 
> Issue https://github.com/netmod-wg/entity/issues/12
> 
>   entConfigChange is useful but maybe covered by netconf-config-change
>   notification, or pub/sub. it is also a bad name, since it fires when
>   the opstate changes...  Need to think about which notifs to define.
> 
> Currently we have three notifications:
> 
>   harwdare-state-change
>   hardware-state-oper-enabled
>   hardware-state-oper-disabled

It seems harwdare-state-change is not netconf-config-change. The
harwdare-state-change indicates that the '/hardware-state/component'
list has changed, i.e., operational state has changed not config.  (It
is always important to read the descriptions and not to draw wrong
conclusions from the fact the SNMP notification was called
entConfigChange.)
 
> Jason Sterne also made the observation on the ML that maybe pub/sub is
> a better mechanism for these simple notifs.

Not sure what 'better' means. I assume that generating these specific
notifications can likely be done way more efficiently than doing the
same via a pub/sub interface unless the pub/sub backend code is really
really smart. That said, I would not be surprised if application
writers will use pub/sub for everything because the hammer makes
everything look like a nail.

/js

-- 
Juergen Schoenwaelder   Jacobs University Bremen gGmbH
Phone: +49 421 200 3587 Campus Ring 1 | 28759 Bremen | Germany
Fax:   +49 421 200 3103 

___
netmod mailing list
netmod@ietf.org
https://www.ietf.org/mailman/listinfo/netmod

[netmod] draft-ietf-netmod-entity issue #12

[Martin Bjorklund]

Hi,

Issue https://github.com/netmod-wg/entity/issues/12

  entConfigChange is useful but maybe covered by netconf-config-change
  notification, or pub/sub. it is also a bad name, since it fires when
  the opstate changes...  Need to think about which notifs to define.

Currently we have three notifications:

  harwdare-state-change
  hardware-state-oper-enabled
  hardware-state-oper-disabled


Jason Sterne also made the observation on the ML that maybe pub/sub is
a better mechanism for these simple notifs.


/martin


___
netmod mailing list
netmod@ietf.org
https://www.ietf.org/mailman/listinfo/netmod