Joel M. Halpern wrote:
To put it differently, the OPS area has as much right to propose their
requirements as any other area (Transport Congestion, Security, ...)
has. And generally, the community has listened to such requests and
gone along with them.
Yes, we have produced a bit of a problem that our initial standards now
have a quality bar comparable with completed work. But we shouldn't
suddenly pick on OPS for that. If we are going to address that problem,
it ought to be in a coherent fashion that discusses how we deal with all
the legitimate requirements, including the fact that stuff has to be
operable.
I agree, and I also agree with Randy about the lack of
standardized NETCONF content. It is a clear indication
that the vendors do not want any standardized configuration
content. Every time 'content' has come up for charter consideration,
the consensus has been to work on something else instead.
But the NETCONF/YANG pieces are coming together, and it will
allow WGs much more power and flexibility than SNMP/SMIv2
to define management interfaces which fit better with
the native CLI than anything the IETF has had before.
That is about as far as OPS-NM area can go.
If a protocol WG do not want to define a
standard configuration interface, then it
does not matter how well SNMP or NETCONF supports
this sort of work.
(One might question whether using the same NM standardization
methodology for 20 years and achieving consistently pathetic
results means the methodology itself needs to be changed,
not just the technology.)
This does not mean we have to simply accept what they (OSP) say. But it
does mean we should give it a fair review, looking at the details,
rather than objecting on principle.
Yours,
Joel M. Halpern
Andy
_______________________________________________
Ietf@xxxxxxxx
https://www.ietf.org/mailman/listinfo/ietf