Re: Management Requirements

rbonica writes:
> Now BEHAVE is asking whether SNMP and IPFIX are the right tools.
> Maybe NETCONF is the right tool for configuration? Maybe SNMP is
> right for fault monitoring? Maybe syslog is right for maintaining a
> record of address mappings? Who knows?

Right implementers have to make many choices which are
predisposed by their personal experience and company leanings.

A given implementation might have the following: an SNMP agent with
particular mib modules and notifications to monitor one or more 
instances of this service. 
IPFIX agent might report complimentary information elements and not necessarily match 1:1 
what one would monitor via SNMP. If configuration is more complex than
can be described in SMIv2 then using YANG to express configuration
makes sense even if NETCONF is never deployed. I find it will end up in the
proprietary CLI/API. And yes syslog gets used for detailed reporting 
diagnostics but have not seen any standard ever need to go that far (yet).

So while not all of these might be used by an operator, having all available in a given
product vastly improves its fit into diverse operational environments.
(Vendor speak for less time to get to full deployment & getting paid).

My advice to product teams has always been to consider the data model in the MIB module
for IETF protocols (or whatever the related mgmt spec for non-IETF technology standards). 
I find that the strings for mgmt terms in these specs get propagated into products doc/interfaces/etc
even if the product doesn't implement an SNMP agent.
SMIv2 syntax MIB modules, especially when written with decent REFERENCE
clauses to the underlying spec, really help to frame which operations and
data should be considered by the implementer using whatever mgmt protocols they choose.


HTH,
Mike MacFaden