Are you using explicit transactions at all? If not, as a quick test, put the _entire_ job in a transaction and see what happens.
-scott On 3/15/07, Ken <[EMAIL PROTECTED]> wrote:
To answer your question: Yes I can use a flat file at this stage, but eventually it needs to be imported into some type of structure. So to that end I decided early on to use sqlite to write the data out. I was hoping for better performance. The raw I/O to read the data and process is around .75 seconds (no write i/o).. So using a flat file output costs about .7 seconds. Using sqlite to do the output costs about 2.25 seconds. My question is why? And what can be done to improve this performance? John Stanton <[EMAIL PROTECTED]> wrote: Ken wrote: > I'm looking for suggestions on improving performance of my sqlite application. > > Here are system timings for a run where the sqlite db has been replaced with a flat file output. > real 0m1.459s > user 0m0.276s > sys 0m0.252s > > This is a run when using sqlite as the output format. > real 0m3.095s > user 0m1.956s > sys 0m0.160s > > As you can see sqlite takes twice as long and almost 8 times the user time. > > Output size for flat file: 13, 360, 504 flatfile.dat > Output size fo sqlit file: 11,042,816 sqlt.db f > > Slite db has the following pragmas set. > > PRAGMA default_synchronous=FULL > PRAGMA temp_store=memory > PRAGMA page_size=4096 > PRAGMA cache_size=2000 > > Any ideas how to get the sqlite output timings to a more respectable level would be appreciated. > > Thanks > Ken > If you want flat file performance, use a flat file. Sqlite is built on top of a flat file and cannot be faster or even as fast. If your application can use a flat file, why use anything more complex? ----------------------------------------------------------------------------- To unsubscribe, send email to [EMAIL PROTECTED] -----------------------------------------------------------------------------
----------------------------------------------------------------------------- To unsubscribe, send email to [EMAIL PROTECTED] -----------------------------------------------------------------------------