Date field performance

Question

hi
The query below works well giving me a few 1000 records in 10-14 secs.
SELECT p1.A AS "Service Number", p.1B AS "Status",
C AS "Suspended Date",
D AS "Reason for suspension"
FROM inst_prod p1, reason p2
WHERE p1.b = 'SUS'
AND p1.reaon_id = p2.reason_cd
But the same query takes 8-10 mins if i add another date condition to it.
SELECT p1.A AS "Service Number", p.1B AS "Status",
C AS "Suspended Date",
D AS "Reason for suspension"
FROM inst_prod p1, reason p2
WHERE p1.b = 'SUS'
AND p1.reaon_id = p2.reason_cd
and trunc(p1.c) between to_date('24-Aug-2006','dd-Mon-yyyy') and
to_date('31-Aug-2006','dd-Mon-yyyy')
Is it because of the TRUNC function ?
Can someone help in fine tuning this query.
Thanks in advance
Vamsi

Answer 1

There is no such issue as "Date field performance". Reason: This implies a consistent and inevitable performance problem when using a date column as oppose to a numeric or varchar column. This is false.
A date column is no slower or no faster than a number or varchar2 column. The speed is determined by what you, the developer or user, does with the contents of that column.
The 1st Step in "Determining The Cause Of A Performance Problem":
Look in the mirror.
The vast majority of performance issues are caused by us, because we are ignorant and doing the wrong thing, misunderstand some basic concept, or worse of all, think we know better and purposefully ignore fundamentals.
Whenever I hit a problem in software engineering, I always blame myself first, before blaming the software. Something like, "you screwed up again Verreynne!". ;-)
It is easy to fall in the trap of "oh, it is not my code, it is because of how the software (Oracle) does this or that (cause performance problems with date fields)".
Don't.
> Is it because of the TRUNC function ?
Yes, that could be a factor. Let's say you have an index on COLUMN_FOO. Data type not important.
So, great, you can write a SQL like this and Oracle will be able to use the index for optimal perfomance:
SELECT * FROM my_table WHERE column_foo = <some-value>
However, what happens to that index when you write the following?:
SELECT * FROM my_table WHERE FUNCTION(column_foo) = <some-value>
The index on COLUMN_FOO is rendered useless. Why? Because the predicate does not compare COLUMN_FOO with some-value. The predicate states that the return value of FUNCTION() needs to be compared to some-value.
So the index cannot be used. Every single row in MY_TABLE has to be read, every single row's COLUMN_FOO has to be run via FUNCTION() and only then can the comparison to some-value be done.
> Can someone help in fine tuning this query.
As others already stated, you need to look at the execution plans of the two queries and determine what the impact is when you add the TRUNC(date) predicate.