大量的更新性能

Question

I have the following update statement below:

UPDATE [dbo].[mytable]
SET [originaldate] =
CASE
    WHEN [datecolumn1] <= [datecolumn2] THEN datecolumn3 
    ELSE [datecolumn1] 
END

This statement executed for almost 5 minutes for 2.6 million of rows. Is there any way to optimize this query?

I tried converting the columns to date like this CONVERT(date, [datecolumn2]) because the columns are datetime but still the execution time is the same.

Also, i tried putting the column and id that I needed in a temp table but still the performance is bad.

Answer 1

Why does it matter if it takes time? This does not seem like something you should be doing regularly.

If it is then you should consider using a computed column

or even in application logic, a view or in the SELECT statement depending on the situation.

Answer 2

Updating all the rows in a table takes time because of logging and transaction overhead. You might find this faster:

select t.*
into #temp
from mytable;

truncate table mytable;

insert into mytable(col1, . . ., originaldate)
    select . . ., 
           (CASE WHEN [datecolumn1] <= [datecolumn2] THEN datecolumn3 
                 ELSE [datecolumn1] 
            END)
    from #temp;