General recommendations before going to changes

Before going to changes, it is essential  to evaluate the figures. What is the total CPU,  hundred seconds/ hour it becomes interesting ... otherwise :
1/ the gain will be so slow so it is not worth doing the changes
2/ the gain will be so slow that the changes effect (statistics figures) will be difficult to detect
3/ Finally, even if the CPU is high, but if it does not happen during your peak 4HRA, you can also forget it , because it will reduce nothing on the IBM bill (for DB2 guys , you can read a good presentation from Phil Grainger about this : "How to save real dollars with SQL tuning")


I know this, but i often forget this rules ....