I think one of the questions that, and perhaps you have, needs to be asked
is: 1) what kind of transaction test produced those numbers? 2) what kind of user load, sim. transactions did the
serverhave? 3) were the transactions non-conflicting? 4) were the transactions read only, read-update, insert? 5)
anda most important question is over what time period was the test taken. This is where various logging, and
checkpointingissues
start to surface. Additionally, some folks do not run the test for any
extended period of time because of memory leakage issues, and other stability issues.
Its very easy to design a test that produces high numbers. Its the
definition of what they mean that causes the problem. Ask the MS
guy to give you the structure of the test to get those numbers. If MS
doesn't release it then they are useless.
As for the TPC tests, they should have the code available. A, and B
are next to useless.
Jordan
----- Original Message -----
From: Jackson, DeJuan <djackson@cpsgroup.com>
To: G. Jayson Stangel <jayson@biztown.com>
Cc: PGSQL Hackers <pgsql-hackers@hub.org>
Sent: Friday, January 08, 1999 1:20 PM
Subject: RE: [HACKERS] ["G. Jayson Stangel" <jayson@biztown.com>] Question
s?
>> Thanks for the info!!! I hope the dentist went well!!!
>It went well, thank you for asking. I got a new dental assistant, and
>she was gentle and sweet (a far improvement). She said that I have good
>teeth, and that I must have been taking good care of them because there
>wasn't much for her to do.
>
>> I have only one question more question(I need to sell the
>> Microsoft guy in the office on this.) I go to Microsoft and
>> they say 20,433 transactions per minute on a quad xenon 450
>> bla, bla, bla. Is there a place where this statistic can be
>> found for Postgre.
>
>I don't know of any such statistics, but maybe some of the other guys
>would.
> -DEJ
>