How to use date column index

Question

Hi,
Why is that I got different results for my queries?
Note that emp_date is indexed, so i dont want to use
a function to it.
QUERY1(full table scan)
SELECT COUNT(*) FROM EMP WHERE TRUNC(EMP_DATE)=TRUNC(SYSDATE-30); (RESULT=8,842)
QUERY2
SELECT COUNT(*) FROM EMP WHERE
EMP_DATE >=(SYSDATE-30) and EMP_DATE < (SYSDATE-30); (RESULT=0)
QUERY3
SELECT COUNT(*) FROM EMP WHERE
EMP_DATE BETWEEN TRUNC(SYSDATE-30) and TRUNC(SYSDATE-30); (RESULT=2,100)
Please help me how to use indexed date column...

Answer 1

> Why is that I got different results for my queries?
Because your queries are different.
1) This one selects all records on 20 october, regardless of their time component
2) This one selects all records with an emp_date at least being 20 october 14:56:30 AND being smaller than 20 october 14:56:30. No record will ever satisfy this condition.
3) This one selects all records on 19./20 october at midnight 00:00:00
Please see this example:
SQL> create table myemp (emp_date)
  2  as
  3   select trunc(sysdate) - 30 +
  4          case
  5          when level <= 2100 then 0
  6          when level <= 8842 then dbms_random.value(1,86399)/86400
  7          else 2
  8          end
  9     from dual
10  connect by level <= 100000
11  /
Tabel is aangemaakt.
SQL> create index i1 on myemp(emp_date)
  2  /
Index is aangemaakt.
SQL> exec dbms_stats.gather_table_stats(user,'myemp',cascade=>true)
PL/SQL-procedure is geslaagd.
SQL> set autotrace on explain
SQL> select count(*) from myemp where trunc(emp_date)=trunc(sysdate-30)
  2  /
                              COUNT(*)
                                  8842
1 rij is geselecteerd.
Uitvoeringspan
   0      SELECT STATEMENT Optimizer=CHOOSE (Cost=88 Card=1 Bytes=8)
   1    0   SORT (AGGREGATE)
   2    1     TABLE ACCESS (FULL) OF 'MYEMP' (Cost=88 Card=1000 Bytes=
          8000)
SQL> select count(*) from myemp where
  2  emp_date >=(sysdate-30) and emp_date < (sysdate-30)
  3  /
                              COUNT(*)
                                     0
1 rij is geselecteerd.
Uitvoeringspan
   0      SELECT STATEMENT Optimizer=CHOOSE (Cost=2 Card=1 Bytes=8)
   1    0   SORT (AGGREGATE)
   2    1     FILTER
   3    2       INDEX (RANGE SCAN) OF 'I1' (NON-UNIQUE) (Cost=4 Card=2
          50 Bytes=2000)
SQL> select count(*) from myemp where
  2  emp_date between trunc(sysdate-30) and trunc(sysdate-30)
  3  /
                              COUNT(*)
                                  2100
1 rij is geselecteerd.
Uitvoeringspan
   0      SELECT STATEMENT Optimizer=CHOOSE (Cost=2 Card=1 Bytes=8)
   1    0   SORT (AGGREGATE)
   2    1     FILTER
   3    2       INDEX (RANGE SCAN) OF 'I1' (NON-UNIQUE) (Cost=4 Card=2
          50 Bytes=2000)
> Please help me how to use indexed date column...
Use this query instead:
SQL> select count(*)
  2    from myemp
  3   where emp_date between trunc(sysdate)-30 and trunc(sysdate)-29 - interval '1' second
  4  /
                              COUNT(*)
                                  8842
1 rij is geselecteerd.
Uitvoeringspan
   0      SELECT STATEMENT Optimizer=CHOOSE (Cost=2 Card=1 Bytes=8)
   1    0   SORT (AGGREGATE)
   2    1     FILTER
   3    2       INDEX (RANGE SCAN) OF 'I1' (NON-UNIQUE) (Cost=4 Card=2
          50 Bytes=2000)Regards,
Rob.