I am suffering performance issues with the EE ODBC stage that I hope you might be able to help me out with. I have had a search, but couldn't find the solution.
Basically, we have several (Parallel) jobs that run sequentially, each of them loading up either sequential files or datasets to MS-SQL tables via the EE ODBC stage. The files are anything up to 50k rows, and the 'widest' one is 413 columns
![Shocked :shock:](./images/smilies/icon_eek.gif)
Now, I have been running tests using cut down files of 500 rows, and have seen some odd degradation of the performance. Taking a single job as an example, if I drop the table, then run a job to create the table and populate it, I get 500 rows/second. If I then amend the job and run as Upsert, the performance falls away dramatically - three runs saw it fall to 167, then 19 and finally 25 rows per second.
Can anyone think of what this might be? Some of the jobs seem to now be running consistently at 250 r/s (although functionality is identical), though earlier runs (prior to dropping the tables) got down to single figures in terms of rows/second.
Thanks in advance for any help that you might be able to offer.
Kind regards,
Stu