Need to know which condition is better

Question

Hi,
i have one table in which most of the records are unique.
What would be better , to create index or to continue without index.
I need faster response of the query . No of rows in table is 32 lac.
below is the output with index and without index for some sample data.
in first scenario i have two indexes.
in second scenario i have only one index
in third one i don't have any index.
here col2 is having only 2-3 distinct values while both other columns are having almost unique values ( different data )
SQL> set autotrace traceonly
SQL> SELECT COUNT(*)
FROM my_table
WHERE col1 =28582
AND col2='1'
AND col3 ='1452631' ;
Elapsed: 00:00:00.47
Execution Plan
Plan hash value: 2123134521
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
| 0 | SELECT STATEMENT | | 1 |16 | 3 (0)| 00:00:01 |
| 1 | SORT AGGREGATE | | 1 |16 || |
|* 2 | INDEX RANGE SCAN| IND_my_table_1 | 1 |16 | 3 (0)| 00:00:01 |
Predicate Information (identified by operation id):
2 - access("col1"=28582 AND "col2"='1' AND
"col3"='1452631')
Statistics
1 recursive calls
0 db block gets
3 consistent gets
0 physical reads
0 redo size
3882329993 bytes sent via SQL*Net to client
492 bytes received via SQL*Net from client
4 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
1 rows processed
SQL> drop index Ind_my_table_1;
Index dropped.
Elapsed: 00:00:00.66
SQL> SELECT COUNT(*)
FROM my_table
WHERE col1 =28582
AND col2='1'
AND col3 ='1452631' ;
Elapsed: 00:00:01.52
Execution Plan
Plan hash value: 2881393183
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
| 0 | SELECT STATEMENT | | 1 | 16 | 4(0)| 00:00:01 |
| 1 | SORT AGGREGATE | | 1 | 16 | | |
|* 2 | TABLE ACCESS BY INDEX ROWID| my_table | 1 | 16 | 4(0)| 00:00:01 |
|* 3 | INDEX RANGE SCAN | IND_my_table_2 | 1 | | 3(0)| 00:00:01 |
Predicate Information (identified by operation id):
2 - filter("col2"='1' AND "col3"='1452631')
3 - access("col1"=28582)
Statistics
199 recursive calls
0 db block gets
31 consistent gets
2 physical reads
0 redo size
514 bytes sent via SQL*Net to client
492 bytes received via SQL*Net from client
4 SQL*Net roundtrips to/from client
6 sorts (memory)
0 sorts (disk)
1 rows processed
SQL> drop index Ind_my_table_1 2;
Index dropped.
Elapsed: 00:00:00.07
SQL> SELECT COUNT(*)
FROM my_table
WHERE col1 =28582
AND col2='1'
AND col3 ='1452631' ;
Elapsed: 00:00:00.70
Execution Plan
Plan hash value: 1679787526
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
| 0 | SELECT STATEMENT | | 1 | 16 | 6431(2)| 00:01:18 |
| 1 | SORT AGGREGATE | | 1 | 16 | | |
|* 2 | TABLE ACCESS FULL| my_table | 1 | 16 | 6431(2)| 00:01:18 |
Predicate Information (identified by operation id):
2 - filter("col2"='1' AND "col1"=28582 AND
"col3"='1452631')
Statistics
169 recursive calls
0 db block gets
29754 consistent gets
0 physical reads
0 redo size
514 bytes sent via SQL*Net to client
492 bytes received via SQL*Net from client
4 SQL*Net roundtrips to/from client
5 sorts (memory)
0 sorts (disk)
1 rows processed
Pl suggest how should i create the index.

Answer 1

Hi,
KriC wrote:
... here col2 is having only 2-3 distinct values while both other columns are having almost unique values ( different data )
SQL> set autotrace traceonly
SQL> SELECT COUNT(*)
FROM my_table
WHERE col1 =28582
AND col2='1'
AND col3 ='1452631' ;
Pl suggest how should i create the index.There's no substitute for actually trying things on your system, with your full data.
You may not be able to try different ideas on your Production system, but your Test system is probably closer that anything that anyone else on this forum has.
My best guess, for queries like the one above, would be a compound index in (col1, col3, col2) or (co3, col1,col2). Col2 is least selective (that is, fewest unique values) so it should definitely be last. There may not be any advantage to including col2 in the index.
Assuming you do create a compound index that starts with col1: that will not help in queries where col1 is not specified, but col3 is. If you sometimes run queries like that, you might want a separate index on col3.