Or you will need to get a faster disk drive! And more of them. Consider a raid 0 system using striping. High speed fiber connects will also be helpful. Can you do a test load to a memory database. What is the performance (row/sec) of your data using an in memory database? Is that a burst rate or a sustained data rate? You need to achieve about 70k rows per second to be able to maintain 66k rps. Maximum TPS is 60 on a 7200 rpm drive. A 15000 RPM drive should get about 120, which is double the TPS. So maybe a faster drive would be helpful. Ken
[EMAIL PROTECTED] wrote: James Steward wrote: > > Michael Ruck wrote: > > > > I know that a natural join exists, but it is not automatic as > > it seems to be in MySQL. > > Thanks , and thanks to all who replied to my questions. > > I've been testing SQLite's speed, for inserting the type of data I > gather from the field. I've tried encapsulating multiple inserts > between begin and commit statements, fiddling the pragmas > and with both Tcl and C interfaces, even using a RAM disk to > store the database file. > > I can receive up to 2,000,000 records in about 30 seconds from > the field, and I can't seem to jam them into an SQLite database > any faster than about 100 seconds at best, on my system. > SQLite will do about 50000 inserts/sec on my Linux workstation. Doing 2 million in 30 seconds is going to be pushing the envelope. You should probably consider a different solution. -- D. Richard Hipp ----------------------------------------------------------------------------- To unsubscribe, send email to [EMAIL PROTECTED] -----------------------------------------------------------------------------
