We're testing following :

DB2/udb version 8 for linux on zseries
mainframe applications are running on VSE/ESA 2.7.1

I'm testing the batch-applications which are connecting to DB2/UDB on linux by using TCP/IP. When we look at the cpu-usage of the program that has been executed i see a cpu-usage X 5 in compare of the program executing it connecting to db2/vse. This increase is big, has anyone experience with this ? One of the causes could be the overhead of DRDA, ideas ?