RE: Objectives discussion

[Andy Bierman <abierman@cisco.com>]

At 09:55 AM 11/20/2001 -0800, Durham, David wrote:
>Hi Andy,
>
>I'm not sure anyone can really construct a high-level opinion w/o a proposal
>to base it on. It would seem that if the cost of change is low, and the
>benefits are sufficiently high, then most people would be in favor of it...
>But let's wait and see if anyone has formulated a high-level opinion at this
>point (I for one found myself answering "it depends" to each of your
>questions).

I think some of the things on the list below are independent
of proposals, but I'm all in favor of more proposals. 
You don't need to see a proposal to know if preserving
SMIv2 features is important to you, or if SPPI is important
to you.

>Also, I would restate number 4 to ask whether we want to continue diverging
>by creating incompatible standards in the IETF, or do we want to attempt
>convergence where possible. 


Of course!, but the phrase "where possible" is quite subjective.

I do not accept the premise that just because some WG went off 
and created a second management data definition language,
almost identical to SMI, and not significantly better than SMI,
that the next generation SMI should be hacked to support it.

We need to distinguish between "better than SMIv2" and merely
"different than SMIv2".  For each feature, if the "SPPI way"
is better than the "SMIv2 way", than it should become the "SMIng way".
If neither way is that good, and somebody comes up with a
better solution, then that should be the "SMIng way".

The argument "we have to do it this way because SMIv2 (or SPPI)
does it this way" is less than compelling. If we can't fix
what's broken then what's the point of this exercise?
This is a standards body, not a Historical Preservation Society.


>Cheers,
>-Dave

Andy


>> -----Original Message-----
>> From: Andy Bierman [mailto:abierman@cisco.com]
>> Sent: Wednesday, November 14, 2001 9:48 AM
>> To: sming@ops.ietf.org
>> Subject: Objectives discussion
>> 
>> 
>> Hi,
>> 
>> I think we need to start analyzing the objectives,
>> starting from a high level. The list of objectives
>> is conflicting in places, and some objectives will
>> cost way more than others to achieve. There are obviously
>> many conflicting viewpoints represented in the objectives,
>> and this is the source of the problem.   
>> 
>> I think there are high-level decisions to make that will help
>> manage all these individual objectives, based on the primary
>> goals of each WG member. It would be nice to know what
>> the WG consensus is on the following -- maybe at the extremes,
>> but likely somewhere in the middle for all of them:
>> 
>> 1) don't want to change anything  ... want to start over
>>   
>> 2) keep the SMI data model ... replace the SMI data model
>> 
>> 3) keep the SMI look-and-feel ... replace the SMI look-and-feel
>> 
>> 4) don't merge SMI and SPPI ... merge SMI and SPPI
>> 
>> 5) don't care about independent protocol mappings ... must support 
>>    independent protocol mappings at any cost
>> 
>> 6) optimize for human usability ... optimize for machine usability
>> 
>> I made it clear where I stand by writing the SMI-DS draft.
>> Hopefully, the rest of the WG will start speaking up as well.
>> 
>> thanks,
>> Andy
>> 
>> 
>>