Some general thoughts: 1) Have you tried loading into the PSA first? 2) Can you expand further on the InfoPack you're using -- I may be able to setup a test on my end... 3) Can you e-mail me any relevant ST03N, ST05 and/or SE30 traces? 4) Are the flat files ASCII or CSV files? SAP recommends ASCII format. FYI: We use Ascential for our ETL loading and don't have this particular issue. -----Original Message----- From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of scott.hutchinson@xxxxxxxxxxxxxxxxxxxxx Sent: Monday, October 10, 2005 4:58 AM To: oracle-l@xxxxxxxxxxxxx Subject: Ridiculously high number of commits All, I have a performance problem while loading data into SAP/BW from flat files. This process for loading the data is a standard SAP routine, and it issues a COMMIT after each record is inserted - we have about 20 million rows to insert, so this is a lot of commits! We've broken the load process into 10 jobs that run concurrently, however they spend the majority of their time sitting around waiting on "log file sync", which is no great surprise. I have a target of 4 hours for loading this data into SAP's "Info Cubes", but this is currently taking 8 hours. Does anyone have any smart ideas for lessening the impact in the database from issuing such a high number of commits? btw - the DB server is a 12 CPU HP running at 12% utilisation during the load. And yes - we are also enganging SAP to see if they can improve their load process. Thanks, Scott Hutchinson Interact Analysis Ltd. ::This message sent using the free Web Mail service from http://TheName.co.uk -- //www.freelists.org/webpage/oracle-l -- //www.freelists.org/webpage/oracle-l