Db2 – Why does “DB2 Recovery Expert” web console doesn’t enable the Run and Next button

If ID1 is more selective than ID2, then construct your index over B with (ID1, ID2, Type_Code) , and make sure there is an index over A by (ID1, ID2).

If ID2 is more selective than ID1, then construct your index over B with (ID2, ID1, Type_Code) , and make sure there is an index over A by (ID2, ID1).

If you are using DB2 for i, then you might also try an Encoded Vector Index [EVI] over B on (Type_Code), in combination with an index by (ID1, ID2) [or the reverse order, as above]. However, the EVI seems unlikely to help in this case.

You won't get any specific answer to this question, because as you observed, it depends.

The answer depends on your company's tolerance for having one DB2 instance adversely impact the performance of another's. You may be able to control or limit this impact if your AIX admins are willing to learn and set up WLM policies at the operating system level (not DB2 WLM).

This is really a question of capacity management. Understanding the workloads in your environment is key here, and being able to identify (or predict) when you'll run into limits will influence your decisions about how many instance(s) to put on a single server.