Very slow distinct

Question

Guys,
Please any of you experts, help me re-write the query.. I see the physical IO seems to be extreamly slow with this query.. I think it may be due to the joins. Is there a better way to re-write this ? I think SORT_UNIQUE is holding up the performance of this query.
SELECT DISTINCT ld.lock_id, ev.entity_id, ev.entity_volume_id,
-- to get new evid
ev.last_updated_by, ev.last_updated_date_time, ev.created_by,
ev.created_date_time
FROM entity_volume ev,
entity_volume_lock evl,
lock_detail ld,
period p
WHERE ev.program_id = :b5
AND ld.program_id = :b4
AND ev.volume_type = :b3
AND ld.period_id = ev.period_id
AND evl.entity_id = ev.entity_id
AND evl.lock_id = ld.lock_id
AND ld.period_id = p.period_id
AND p.month_year BETWEEN :b2 AND :b1
Explain plan is as below
Operation     Object Name     Rows     Bytes     Cost     Object Node     In/Out     PStart     PStop
SELECT STATEMENT Optimizer Mode=ALL_ROWS          7           1993                     
SORT UNIQUE          7      679      1993                     
FILTER                                        
TABLE ACCESS BY LOCAL INDEX ROWID     ENTITY_VOLUME     1      61      3                     
NESTED LOOPS          7      679      1970                     
NESTED LOOPS          553      19 K     311                     
NESTED LOOPS          135      3 K     41                     
TABLE ACCESS FULL     PERIOD     1      11      2                     
INDEX RANGE SCAN     LOCK_DETAIL_IDX002     1      15      39                     
INDEX RANGE SCAN     ENTITY_VOLUME_LOCK_IDX_002     41      410      2                     
PARTITION RANGE SINGLE                                   KEY     KEY
INDEX RANGE SCAN     ENTITY_VOLUME_IDX_001     1           2                KEY     KEY

Answer 1

I'm not certain why Mr. Beilstein never replied and I believe his claim that "a programmer can almost always bypass the need to do a distinct" may be a little optomistic.
However, consider the two queries below:
SELECT CBONAME CBONAME
FROM (
SELECT 1 ORDNUM, 'MAN_'||TO_CHAR(SYSDATE,'YYYYMMDD')||UPPER('_'||:STRUSER) CBONAME FROM DUAL
UNION ALL
SELECT DISTINCT 2 ORDNUM, BATCHNAME CBONAME FROM LND_STAGESALESTRANSACTION
WHERE UPPER(BATCHNAME) LIKE UPPER('%_'||:STRUSER) OR :ISADMIN = 1
AND SUBSTR(UPPER(BATCHNAME),1,4)='MAN_'
AND (STAGEPROCESSFLAG = 0 OR STAGEPROCESSFLAG = :SHOWPROCESSED)
ORDER BY ORDNUM
SELECT CBONAME CBONAME, SUBSTR(CBONAME,LENGTH(:STRUSER)*-1)
FROM (
SELECT 1 ORDNUM, 'MAN_'||TO_CHAR(SYSDATE,'YYYYMMDD')||UPPER('_'||:STRUSER) CBONAME FROM DUAL
UNION ALL
SELECT 2 ORDNUM, BATCHNAME CBONAME FROM LND_STAGESALESTRANSACTION
WHERE SUBSTR(UPPER(BATCHNAME),1,4)='MAN_'
AND UPPER(SUBSTR(BATCHNAME,LENGTH(:STRUSER)*-1)) = UPPER(:STRUSER) OR :ISADMIN = 1
AND (STAGEPROCESSFLAG = 0 OR STAGEPROCESSFLAG = :SHOWPROCESSED)
GROUP BY BATCHNAME
The first is an example using distinct and like clauses. The second uses group by and substr. The second is twice as fast as the first.