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 8:44 AM
Subject: Re: action RPC I-D
...
If one knows a system's initial configuration, and has a sequence
of successful edit-config operations, is it possible to compute
the final configuration state (in order to support, for example,
disaster recovery) without knowing the internals of the system
being configured?
Replay on the same box with the same conditions should produce
the same results. Replay across arbitrary boxes from different
vendors? It depends on the RPCs and the data models.
That's the question I'm asking. Let me break it into some simpler
questions:
1) if one successfully plays the same sequence of commands
into two different boxes with the same starting configuration
(not asking about whether the vendors are different, just whether
the boxes' configurations are identical) can one assume that
the resulting configurations will be the same?
2) is the result (in terms of how the system's configuration will change)
of a successfully completed RPC deterministic?
3) is the result (in terms of the changes to a system's configuration)
of a successfully completed RPC predictable?
If the answer is "no" to any of these questions, then I think we're in
for a tough time explaining how this belongs in "configuration management",
and need to face up to what this protocol really is.