Subject understanding slow commit
Author svanderclock
hello,

i m still with my "slow" commit. i have one question, when i add some row with some key, the commit is very fast, but when i add other row with other key the commit is very very slow

for exemple

delete from xxx where ID_CUST = 'sMszqqqIN';
Insert into xxx (.. ID_CUST ..) value (... 'sMszqqqIN' ...)
Time to execute the sql: 125 ms
Indexed Read: 101
Non Indexed Read: 0
Inserts: 1
Updates: 0
Deletes: 1
Time to commit the data: 14 ms


delete from xxx where ID_CUST = 'ADsMszqqqIN';
Insert into xxx (.. ID_CUST ..) value (... 'ADsMszqqqIN' ...)
Time to execute the sql: 115 ms
Indexed Read: 101
Non Indexed Read: 0
Inserts: 1
Updates: 0
Deletes: 1
Time to commit the data: 814 ms

ID_CUST is part of multiple "multi column INDEX"

why some time, when just the value of one column change, the commit can be 100x more slower? when i read the task manager, during the commit, nothing i read from the harddrive, only written to it.

is their is a way to improuve it (exept set force write off)

thanks by advance
stephane