On Fri, Aug 20, 2010 at 09:59:12PM +0200, Thomas Narten wrote:
 
> Thanks for bringing up this topic!
> 
> > OLD
> 
> >    Network Management MAY be supported by IPv6 nodes.  However, for IPv6
> >    nodes that are embedded devices, network management may be the only
> >    possible way of controlling these nodes.
> 
> > NEW
> 
> >    IPv6 nodes may support network management via the Simple Network
> >    Management Protocol version 3 (SNMPv3) [RFC3411].  For IPv6 nodes
> >    that are embedded devices, network management via SNMPv3 may be the
> >    only possible way of controlling these nodes.
> 
> > Rationale:
> 
> >    I think it is important to provide a explicit references to SNMPv3
> >    to make it clear that SNMPv3 is the SNMP version to be
> >    implemented.
> 
> I'm fine with the reference to SNMPv3.
> 
> But, is that last sentence really accurate? Are embedded devices
> typically or normally managed via SNMP?
> 
> And is there even a generally accepted definition of what an embedded
> device is? Is it the same as, say, a network appliance?

The sentence in question was there before - I just removed the
'However'. Since the sentence neither uses 'typically' nor 'normally',
I am not sure I understand your first question. If your proposal
behind your questions is to remove the sentence, I would not object.
 
> > OLD
> 
> >    IP Forwarding Table MIB [RFC4292] SHOULD be supported by nodes that
> >    support an SNMP agent.
> 
> > NEW
> 
> >    The IP-FORWARD-MIB module [RFC4292] SHOULD be supported by nodes that
> >    support an SNMP agent.
> 
> > Rationale
> 
> >    I believe it is clearer to list the MIB module name instead of a
> >    paraphrase of the respective RFC title.
> 
> Agreed.
> 
> But thinking a bit, should we even go so far as saying SHOULD for the
> two above MIBs when SNMP is implemented?

Two? You include the IP-MIB in the following comment as well?

> The reason I ask is that a year or two ago, when I looked into the
> question of what MIBs to implement for hosts, my conclusion was that
> MIBs aren't really used in the host/server space much. They are
> clearly used for monitoring (but not necessarily configuration) in
> routers.
> 
> I guess I'm asking should we give the same recomendation for all
> devices, or should the recommendation for routers be different from
> that of hosts?

The question is whether we expect network management applications to
look at IP forwarding tables of hosts or not. I do not have a good
answer for this. The ipForwardReadOnlyCompliance allows read-only
implementations so monitoring vs. configuration is not an issue I
think. We could even be more explicit and state that explicitly by
refering to the compliance statement.

I believe the IP-MIB with read-only compliance SHOULD be on every node
that can afford it. On not resource constrained systems, I do not
recall having seen an SNMP agent that does not implement the old
IP-MIB plus the old IP-FORWARDING-MIB. My feeling is that if IPv4
nodes support SNMP, they do support the basic MIB modules we have.

/js

-- 
Juergen Schoenwaelder           Jacobs University Bremen gGmbH
Phone: +49 421 200 3587         Campus Ring 1, 28759 Bremen, Germany
Fax:   +49 421 200 3103         <http://www.jacobs-university.de/>
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to