Hi, I have a table with 2 indexes, the first one defined like:

column1, column2, column3

and the second one

column1, column2

I am not sure if a can eliminate index 2. I supouse that, if index 1 has column1 and column2, maybe DB2 could use index one to cover both cases, I mean, on a where clause, DB2 has to match the exact columns of an index to use it or just the first columns ?

Thanks

You should definitely drop the second index, it is redundent. If the second index was on (column2, column3) that would be different.

The second index is only redundant, if it is not unique, but if it is unique, you will loose the uniqueness on these two columns. So, may be the first index is redundant or only makes sense, if there are queries, where the access path becomes index only.

You're right, but it's also possible to include columns in the index which are not part of the index key.

CREATE UNIQUE INDEX INDEX1 on TABLE1 (COLUMN1, COLUMN2) INCLUDE (COLUMN3)

which makes the second index redundant (this might be different on DB2 OS/390)

Yes this is different for DB2 z/OS. Even in Version 8 that will not be supported.