Wow, thanks for such a quick response ! >Disabling the journal would roughly half the amount >of disk (flash) I/O required. If it currently takes >2 minutes to delete, disabling the journal would >reduce that to about 1 minutes. Still a long time.
Ok, thanks, that's good to know. >Do you also understand that if you power-off during >the update, or if your program crashes for any reason, >the *entire* database file will be irrecoverably >corrupted if you do not have a journal? Yes I do. This is not as critical on our system as it probably is on most others. The database on our embedded system is on battery backup and initialized by a host controller. So if a database goes bad on our embedded system, an error message is sent to the host and the host re-initializes the embedded system, including database files etc. So, we're prepared to deal with database corruption should it happen (very rarely we hope), but for normal operations we'd be willing to risk database corruption for better performance in *some* cases. But, it sounds like suppressing journal file creation won't give us a drastic performance improvement so we'll have think of something else. Thanks again for your response. -Randy
