加载数据或更新列 Oracle-sql

Question

I have a table with a very large volume (oracle 11g), and I do thousands of updates but this in terms of performance takes hours. Some fields and columns from my table:

DEP     EMP FA  LO  DAC
98001   333 123 124 null
98001   333 321 132 null
98001   333 234 233 null

Here is the 1st method (query executed via SQL plus):

WHENEVER SQLERROR EXIT SQL.SQLCODE
UPDATE My_table set DAC= '5' where DEP ='98001' AND  EMP ='333' and FA='123' and LO='124'  
UPDATE My_table set DAC= '8' where DEP ='98001' AND  EMP ='333' and FA='321' and LO='132'  
UPDATE My_table set DAC= '9' where DEP ='98001' AND  EMP ='333' and FA='123' and LO='233'  
… around 10 000 lines
Commit ;

2nd method (query executed via SQL plus):

UPDATE My_table
SET DAC = ( 
CASE  
when FA='123' and LO='124'  THEN  '5'
when FA='321' and LO='132'  THEN  '8'
when FA='234' and LO='233'  THEN  '9'
… around 10 000 lines
END) where DEP ='98001' AND  EMP ='333'
AND FA IN ('123', '321', '234', …. around 10 000 FA);

Is there a faster way to load sql the field? via loading only the field according to the condition of the where clause? Or insert my field? … Thanks for your help.

Answer 1

1st method

Create one index on multiple columns in your big fat table

create index <index_name> on <table_name> (DEP, EMP, FA, LO);

2nd method

While I doubt that it will solve your performance issue, there is a workaround for the max 1000 IN parameters.

You can work your way around it by making it a 2 value comparison like this

UPDATE My_table
SET    DAC = ...
WHERE  DEP ='98001'
AND    EMP ='333'
AND    ('dummy',FA) IN (('dummy','123'),('dummy', '321'), ('dummy','234'), …. around 10 000 FA);

That way you have a maximum of 100,000 values.

Another tip

Check the datatype of your DEP, EMP, FA and LO columns, you might be doing VARCHAR2 vs NUMBER comparisons. And though it works, oracle does execute an implicit conversion to be able to do a compare, which can cause performance issues when comparing lots and lots of values.