RE: Some new results

["Joe Pluta" <joepluta@xxxxxxxxxxxxxxxxx>]

> From: Scott Klement
> 
> I read a bunch of the posts. Some observations:

Scott, two things:

1. If Paul doesn't want to post his observations here (and thus be
available to back them up), I'm not sure you posting them here does any
good except to further muddy the waters.

2. Based on Paul's statements, he clearly didn't read the thread well
enough to understand what I was trying to simulate.  And his conclusions
are bizarre, to say the least:  "A set could be zero or one records".
Pfui.  The results CLEARLY show that using SQL for single record I/O is
idiotic, no matter what Paul says.

However, I'm NOT going to get into a third party debate, which is why I
find your posting of his comments to be inappropriate.  If Paul wants to
come here and debate, or go to the IAAI and debate, fine.  Until then,
it's just dueling quotes and those don't do anyone any good.

For example, here's another quote from Paul, and this one is public: "My
general assessment is that the performance difference between a
well-designed and well-tuned SQL application and a well- designed and
well-tuned traditional application is rarely significant enough to
determine which approach to use across the board."

He provides no figures, just his assessment - the assessment of a man
who sells SQL for a living.  And since the numbers have shown clearly
that SQL really sucks in some basic business situations, it's also
pretty clear to me how valid that assessment is.

Joe