Question

Friends
while executing where clause in Oracle SQL suppose I have

UPDATE schema1.TBL_SCHEMA1_PROCESS_FEED F
                 SET F.TBL_SCHEMA1_PROCESS_LINE_ID      = V_LINE_ID,
                     F.TBL_SCHEMA1_PROCESS_LINE_TYPE_ID = V_LINE_TYPE_ID,
                     F.TBL_SCHEMA1_PROCESS_LINE_SUB_TYPE_ID = V_SUB_TYPE_ID,
               WHERE F.CURR_DATE = V_CURR_DATE  
               AND   F.NEXT_DATE = V_NEXT_BUSINESS_DATE OR F.NEXT_DATE IS NULL;

How this code can be optimized for the condition

F.NEXT_DATE = V_NEXT_BUSINESS_DATE OR F.NEXT_DATE IS NULL

Answer 1

Is that your actual where clause? Do you mean it to be:

WHERE F.CURR_DATE = V_CURR_DATE  
  AND ( F.NEXT_DATE = V_NEXT_BUSINESS_DATE 
        OR F.NEXT_DATE IS NULL )

If so then you need an index, unique if possible, on curr_date.

If you’re not satisfied that this provides a large enough improvement in the execution time then think about extending it to curr_date, next_date. Don’t create a larger index if you don’t need to.

You might also consider chaning your conditions slightly, though I doubt it would make much, if any, difference.

WHERE F.CURR_DATE = V_CURR_DATE  
  AND NVL(F.NEXT_DATE, V_NEXT_BUSINESS_DATE) = V_NEXT_BUSINESS_DATE 

The best possible option is, to update using the rowid. Without a lot more information it’s impossible to know if you’re in a situation where this might be possible but as the rowid is a unique address in the table it always is quicker than indexes, when updating a single row. If you’re collecting data from this table then populating your variables before writing back to the table then this would be possible.

Are those your actual schema and table names… if they are then why not think about chosing something more descriptive?