Re: configuration: writable MIB modules versus NETCONF/YANGmodules

Thomas Nadeau <[email protected]> Thu, 20 Feb 2014 09:50:05 -0500
Newsgroups	gmane.ietf.ops
Message-ID	<[email protected]>
On Feb 20, 2014:4:37 AM, at 4:37 AM, t.petch <[email protected]> wrote:

> ----- Original Message -----
> From: "Warren Kumari" <[email protected]>
> To: "Benoit Claise" <[email protected]>
> Cc: "t.petch" <[email protected]>; <[email protected]>
> Sent: Tuesday, February 18, 2014 10:08 PM
>> On Mon, Feb 17, 2014 at 7:25 PM, Benoit Claise <[email protected]>
> wrote:
>>> Hi Tom,
>>> 
>>> The important questions are:
>>> - will the writable aspects of the MIB module be implemented by the
> vendors
>>> (basically requested by the operators)?
>> 
>> Who knows if it would be implemented by the vendors, but I'm sure most
>> operators would refuse to use it.
>> As a test I just proposed enabling SNMP write and got some good
>> giggles (actually I got a "Good plan! Do  it with snmpv1, no one uses
>> that.", but I'm not really able to convey the sarcasm / mocking tone
>> in mail).
>> 
>> I know of no large operator / network that allows write access -- I
>> suspect that the risks are not really as bad as folk make out, but it
>> has become network lore that snmp write is bad mojo and once something
>> becomes lore...
> 
> I say again, go look at the mpls list, where
> draft-ietf-mpls-tp-te-mib
> has been returned to the WG to consider the question of making it
> read-only, and several have spoken up against.  They are of course
> speaking as individuals and are not speaking for operators, or
> manufacturers, but they are a constituency in favour of writable MIB
> modules.
> 
> And yes, I have in the past worked with enterprises who configure with
> SNMP (with security provided other than by SNMP).

	How many of those few enterprises that enable snmp writes do you think run MPLS-TP? 

	--Tom


> 
> Tom Petch
> 
>> Things may be different in the enterprise space (I think ciscoworks
>> tries to use snmp write). Oh, and Bay Network's Site Mangler used to
>> -- which is why most people used BCC...
>> 
>> W
>> 
>> 
>> 
>>> - will the writable aspects be used by the NMS'?
>>> 
>>> Regards, Benoit
>>> 
>>>> ----- Original Message -----
>>>> From: "Benoit Claise" <[email protected]>
>>>> To: <[email protected]>
>>>> Sent: Friday, February 14, 2014 12:39 AM
>>>>> 
>>>>> Dear all,
>>>>> 
>>>>> We occasionally see read-write MIB module proposals within the
> IETF.
>>>>> However, the write capabilities of those MIB modules are rarely
>>>> 
>>>> implemented.
>>>>> 
>>>>> While discussing this issue with the MIB doctors, we arrive to the
>>>>> conclusion that it's now time to set the direction for future MIB
>>>>> developments within the IETF. Basically, let's not specify
> read-write
>>>>> MIB modules unless we have a good reason. Read-only MIB modules
> are
>>>>> still fine though, as SNMP is clearly used for monitoring
> purposes.
>>>>> 
>>>>> Here is the statement we came up with:
>>>>> 
>>>>>     The OPS area recommends the use of NETCONF/YANG standards for
>>>>>     configuration. IETF working groups are therefore encouraged
> to use
>>>>>     the NETCONF/YANG standards for configuration, specifically in
> new
>>>>>     charters. SNMP MIB modules modifying persistent configuration
>>>> 
>>>> state
>>>>> 
>>>>>     should only be produced by working groups in cases of clear
>>>> 
>>>> utility
>>>>> 
>>>>>     and consensus to use SNMP write operations for configuration.
>>>>> 
>>>>> Ideally, this should become an IESG statement.
>>>>> Your feedback is most welcome.
>>>> 
>>>> You probably know that there is currently a consensus call in
> progress
>>>> over
>>>> 
>>>> "Please indicate Support or Oppose for this MIB
>>>> (draft-ietf-mpls-tp-te-mib) to be read-only."
>>>> 
>>>> and I was pleasantly surprised to see opposition, albeit with one
>>>> opponent later changing their mind (perhaps they were sat on from a
>>>> height:-).  I don't see much sign of Netxxxx in the mpls arena but
> there
>>>> is a need there for configuration, since there need not be a
> control
>>>> plane to set things up.  Perhaps writable MIB modules will still be
> with
>>>> us in the distant future in some corners of the Internet..
>>>> 
>>>> Ideally, this idea would be fleshed out in an I-D.
>>>> 
>>>> Tom Petch
>>>> 
>>>>> Regards, the MIB doctors & Benoit
>>>> 
>>>> .
>>>> 
>>> 
>>> _______________________________________________
>>> OPS-AREA mailing list
>>> [email protected]
>>> https://www.ietf.org/mailman/listinfo/ops-area
>> 
> 
> _______________________________________________
> OPS-AREA mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/ops-area
>

_______________________________________________
OPS-AREA mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ops-area
signature.asc (application/pgp-signature, 842 B)