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

Re: action RPC I-D

Randy Presuhn wrote:

Hi -


From: "Andy Bierman" <ietf@andybierman.com>
To: "Randy Presuhn" <randy_presuhn@mindspring.com>
Cc: "'Netconf (E-mail)'" <netconf@ops.ietf.org>
Sent: Wednesday, November 01, 2006 1:53 PM
Subject: Re: action RPC I-D

...

Seems like the quality of the RPC or data model definition
and documentation would determine how predictable the result
would be on any given system.  No different than SNMP here.
A bad MIB design does not prove that SNMP is broken either.

...

We're not talking about SNMP.  Were talking netconf, a protocol that
was supposed to better support configuration management than SNMP.

With this kind of behaviour, what's the point of tracking configurations?
It seems the only workable solution if equipment can be this arbitrary
will be to tweak things until the device works, and then do a get-config
to figure out what the configuration really is, and hope that none of
the RPCs did things that won't be reflected in the results of get-config,
and hope that replaying the results of the get-config back into a device at
some time in the future will result in the desired configuration.

This seems somewhat suboptimal.


I don't understand your objection.
We should assume that programmers use RPC methods with
an understanding of what each one does.

This is no different than understanding every single data model
and knowing what happens when it is changed with edit-config.

I would assume that a standard RPC method template would
include details like how any portion of the configuration
database is affected (if at all) by the RPC.

I think it is presumptive to assume that proper NE device management
requires all configuration procedures to be modeled as data, which can
be editing in almost arbitrary ways, as if if was nothing more than
a text file on the managed device.   IMO this approach has failed
spectacularly, and it's time to try something else.





Randy


Andy

--
to unsubscribe send a message to netconf-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/netconf/>