Hi. We run 2012 enterprise. I just finished a poc that showed if I insert my finder keys to db2 thus allowing my joins on db2 to render 10's of thousands of lookup records instead of 10's of millions, that my job can run in about half the time. The run time reduced from 21 minutes to 10. The db2 datasets are used in merge joins for lookup purposes. SQL Server datasets serve as the left side of most of these merge joins. In one case a sql server dataset is inner joined with the db2 data.
I noticed that the finder inserts themselves were taking a substantial amount of that 10 minutes. Anywhere from 3-6 minutes. And the source sql server query was only 8 seconds of that 3-6 minutes in the insert DF. There are only 46k records that need to be inserted into the db2 finder table.
My db2 connection says my provider is Native OLE DB\IBM OLE DB Provider for DB2 - DB2COPY1. My db2 table name is from a variable because for some reason the dropdown of db2 table names didn't seem to go far enough into the alphabet for me to pick my table.
I poked around and concluded that while fastload rowset and fastload rowset from variable could be chosen in the properties of the target oledb db2 component, choosing them over "rowset from var" wasn't legal and in fact turned the destination component red.
Does the community have any recommendations for making these inserts perform faster?