Hi, This is a guess, but I would assume that would wouldn't get accurate result by using ODBC.
If you write your test in C, you may as well use the C library available and not use ODBC. Just my 0.02$ worth. Daniel > > Hi. > > I believe you missed his point: He expected a time difference but did > not see any (statistically significant) differences. > > A try on the original question: > > Yes, adding the command line option flush in the appropriate my.cnf > sections is a correct way to enable the "flush mode". > > Why you see hardly a difference, I am not sure. > > Anyhow, 1000 inserts in 0.2 seconds is about 5000 inserts per second. > That is much too high for forced disk writes after each insert, except > if you have a big RAID array, IMHO. > > I played around a bit on my testserver, a linux box with a some 2.2 > kernel (got about 2500 inserts per second), but also got no difference > between flush on vs. off. A look at vmstat suggests, that the data is > cached and written with a delay. The delay is bigger without flush, > but also existent with flush. Also, the number of written blocks had > to be a lot bigger with flush enabled (at least 1000), but isn't. > > Therefore, it seems that in my environment, an additional layer is > caching which cannot be influenced by MySQL (via the means of > fsync)?!? > > Bye, > > Benjamin. > > On Fri, Sep 28, 2001 at 03:29:34PM -0600, [EMAIL PROTECTED] wrote: > > if it flushes it has to do more disk writes, which means it has to wait > > for I/O to complete and it slows down. If hard drives were the > same speed > > as RAM we wouldn't have these problems :) > > > > ----------- > > Spinlock > > > > Creator of EmpireQuest > > http://www.empirequest.com > > > > On Thu, 27 Sep 2001, Meenatchi Anbuchelvan wrote: > > > > > Hi > > > > > > Iam little new to MySQL. I have done a simple coding > > > in C using ODBC API, which inserts say, 1000 records. > > > The way i have done it is: loop thro 1000, insert a > > > record for each iteration. Basically I want to see how > > > fast the inserts are with and with out flush. I > > > started the mysqld in flush mode by adding an option > > > "flush" under the group "mysqld" in the file > > > /etc/my.cnf.( is the right way to start mysqld in > > > flush mode?, the other way of starting from command > > > line using flush mode didnt work for me) When I > > > checked the variables, I saw flush was ON. I executed > > > my script and saw total running time close to > > > 0.200-0.217 secs. I got this for several runs. Then I > > > removed "flush" from config file and started the > > > daemon. This time I expected to see much lesser total > > > running times. However I see times in the same range > > > as before. > > > > > > When I tried increasing the number of records, I saw > > > time= 1.033 sec for flush mode and time in range of > > > 1.000 sec for non-flush mode. > > > > > > Iam very confused. COuld anyone explain. > > > -- > [EMAIL PROTECTED] > > --------------------------------------------------------------------- > Before posting, please check: > http://www.mysql.com/manual.php (the manual) > http://lists.mysql.com/ (the list archive) > > To request this thread, e-mail <[EMAIL PROTECTED]> > To unsubscribe, e-mail > <[EMAIL PROTECTED]> > Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php > --------------------------------------------------------------------- Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail <[EMAIL PROTECTED]> To unsubscribe, e-mail <[EMAIL PROTECTED]> Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
