Re: DB Buffer Cache Size

On Wed, 25 Aug 2004 23:00:50 +1000, "Howard J. Rogers" <hjr_at_dizwell.com> wrote:

>Ed Stevens wrote:
>
>> On Wed, 25 Aug 2004 07:46:27 +1000, "Howard J. Rogers"
>> <hjr_at_dizwell.com> wrote:
>>
>> <snippage>
>>>On the other hand, a DBA or manager needs to have yardsticks by which to
>>>judge or assess a product, even if there's nothing much beyond that he can
>>>do. When you are evaluating two potential competitor products, . . . "
>>>
>> Who gets to evaluate two potential competitor products?
>
>Lots of people. Lots of times. Me included, in the dim and distant past.
>
>In a world of mere anecdote, we can bat this sort of thing back and forth
>until we're exhausted!
>
>The principle remains: you can't make sensible judgements about even a
>solitary piece of shrinkwrap without a sound starting point, which has to
>be a judgement of design and coding efficiency.
>

Agreed. I was simply being a bit wistful at the thought of there even being a technical evaluation of shrink-wrap. Or if there were, it being more than a pro forma justification for a decision that was already cast in concrete. I know other's experiences must be different than mine.

>> The only
>> experience I've ever had with shrink-wrap is "The user department has
>> purchased this. Make it work." Or the slight variation, "The user
>> department has purchased this. The vendor has already installed and
>> configured the server, the db, and the app. Make it work."
>
>Yeah, I've been there too. And when I couldn't make it work, what did I do?
>Collected a lot of information about *why* it wouldn't work (actually, the
>example I have in mind had nothing to do with Oracle and everything to do
>with not supplying 3 crucial DLLs, and a further 4 which conflicted with
>practically every other DLL that had ever been written by anybody
>anywhere... but the thing still came down to an assessment of their design
>and the implementation of their design).
>
>It's either that, or look like a complete prat when the inevitable question
>arises "Why doesn't this work" and one's sole response is "I dunno".
>
>Regards
>HJR
>
>> The closest I've *ever* been to evaluating a product was sitting in a
>> meeting where the proposed vendor was asking about moving the data
>> from Oracle to Access for reporting because "we've found that Oracle
>> slows to a crawl with more than ten concurrent connections." You can
>> draw your own conclusions.
>>
>> <snippage>
>
>
Received on Wed Aug 25 2004 - 08:22:13 CDT