[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: snmpconf Status of our work



Hi,

I won't challenge the advancement of the document, but will respond to
your response.

To suggest that this suggestion is "late in the process" is a bit
misleading. I have been requesting the split of the snmpconf
functionality for approximately three years, so we could reuse existing
mib module functionality where possible, especially in terms of
scripting support, and so that other working groups could more easily
reuse some of the work being done in snmpconf. 
I think it is too bad when "late in the process" is used to justify
pushing through work that is technically not as well-designed for
reusability as it should be for an industry  standard. 

You comment that it would take a significant effort to do the split. It
will require even more effort to split it later, since that would almost
certainly require re-rooting the tree, which of course requires
obsoleting and renaming descriptors and reassigning OIDs.

Time-to-market is certainly a reality to consider, but what vendors do
you know are just waiting to implement this specification once it
reaches PS? I haven't heard anybody even mention implementing snmpconf
in the past year or so. 

The only places I have seen snmpconf mentioned is in MIDCOM, where they
have largely rejected using it because it is too heavyweight, especially
with the snmpconf scripting. If the document were split into the script
language, script execution support, and policy support, the MIDCOM WG
(and probably others) would be much more likely to utilize some of the
snmpconf functionality.

My $.02 and now I'll shut up,
dbh

> -----Original Message-----
> From: David Partain (LI/EAB) [mailto:david.partain@ericsson.com] 
> Sent: Tuesday, October 21, 2003 7:07 AM
> To: snmpconf@snmp.com
> Subject: Re: snmpconf Status of our work
> 
> 
> Hi all,
> 
> On Friday 17 October 2003 16.52, Harrington, David wrote:
> > I continue to believe that the scripting language used to 
> specify policies
> > should be separated from the MIBs that coordinate the 
> advertisement and
> > coordination of policies.
> [rest deleted]
> 
> Thanks for the mail and the comments.
> 
> It seems very late in the game to be suggesting a split in
> the architecture of the document.  This would definitely be a
> very major change at a time when our only remaining milestone
> is to close the working group.
> 
> While there may be reasons to consider a split of the MIB
> constructs and the scripting language, it would require a
> significant effort to accomplish this.  Our preference would
> be to get the document published as is to allow vendors who are
> interested in implementing the work to do so.  Should we indeed
> discover wide support for a looser coupling between scripting
> and the MIB module, we should clearly address that in future
> work in a rejuvenated SNMPCONF or in some other working group.
> 
> I hope that seems reasonable...
> 
> Cheers,
> 
> David Partain
> 
>