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

Re: Diffs in the Hardaker and Perkins Drafts



HI,

Below is one of the fundamental differences....
At 04:23 PM 9/23/2002 -0700, Wes Hardaker wrote:
>>>>>> On Mon, 23 Sep 2002 19:07:57 -0400, "Harrington, David" <dbh@enterasys.com> said:
>
>David> I believe Dave Perkins' point that columnar retrieval is
>David> necessary is important. Few applications actually focus on
>David> retrieving complete tables, but many focus on retrieving
>David> certain columns.
>
>Actually, I think most people are in agreement on that subject.  David
>P. and I still have disagreements (which we hammered out on the phone
>the other day as a disagreement and made sure it wasn't a lack of
>understanding each other) as to whether it should be possible to
>request all columns of a table or not (in David P.'s eyes, you should
>always be forced to specify a list of what columns you want and the
>protocol should never allow you to say "return them all").
And the reason is that, since a MIB may be extended, then what worked
yesterday (before the extension), might not work today. (Worked means
met your performance and network utilization goals. Extra data (from
those added columns) is not used by your app, and does nothing but
slow down obtaining the data you do want.

The other point was that I wanted to be able to specify columns from
related tables in the same request. For example, columns from the
IF table and the Ethernet extension table (filter on ifType = Ethernet).
Wes's proposal didn't support this because it allowed the iteration
order to be determined by the agent on each table, and, thus, the
iterration order for the IF table and the Ethernet extensions table
could be different. When different, it would be too expensive for the
agent to do a "join" to put them back together to return a row in the
resulting "projection". 

I think we both came up with reasonable results given different
assumptions.

Regards,
/david t. perkins