How to make the sql with unions to run faster

Question

I have the following sql which takes long time to run in sqlplus. This is in oracle 10, can I use hints to speed up the process. Union all brings all rows, but I want to eliminate the duplicates.
select A.EMPLID,S.OPRID,A.LASTUPDDTTM,A.EMPL_RCD_NBR,A.LAST_NAME,A.FIRST_NAME,A.DEPTID,B.COMPANY,B.PAYGROUP,B.PAY_END_DT,B.OFF_CYCLE,
B.ERNCD_REG_HRS,B.REG_HRS,0,B.REG_HRLY_EARNS
from PS_EMPLOYEES A,PS_PAY_EARNINGS B,PS_SCRTY_TBL_DEPT S,PSTREENODE T
where S.ACCESS_CD = 'Y' and T.SETID = ' ' and T.TREE_NAME = 'DEPT_SECURITY'
and T.EFFDT = S.TREE_EFFDT and T.TREE_NODE = A.DEPTID and T.TREE_NODE_NUM between S.TREE_NODE_NUM
and S.TREE_NODE_NUM_END
and not exists (select 'X' from PS_SCRTY_TBL_DEPT S1
where S.OPRID = S1.OPRID
and S.TREE_NODE_NUM <> S1.TREE_NODE_NUM
and T.TREE_NODE_NUM between S1.TREE_NODE_NUM
and S1.TREE_NODE_NUM_END
and S1.TREE_NODE_NUM between S.TREE_NODE_NUM
and S.TREE_NODE_NUM_END)
and B.EMPLID = A.EMPLID and B.EMPL_RCD_NBR = A.EMPL_RCD_NBR
and B.SINGLE_CHECK_USE IN ('C', 'N') and (B.REG_HRS<>0 or B.REG_HRLY_EARNS<>0)
and B.PAY_LINE_STATUS in ('C','F')
union
select A.EMPLID,S.OPRID,A.LASTUPDDTTM,A.EMPL_RCD_NBR,A.LAST_NAME,A.FIRST_NAME,A.DEPTID,B.COMPANY,B.PAYGROUP,B.PAY_END_DT,B.OFF_CYCLE,
B.ERNCD_REG_HRS,B.REG_HRS,0,B.REG_EARNS
from PS_EMPLOYEES A,PS_PAY_EARNINGS B,PS_SCRTY_TBL_DEPT S,PSTREENODE T
where S.ACCESS_CD = 'Y' and T.SETID = ' ' and T.TREE_NAME = 'DEPT_SECURITY'
and T.EFFDT = S.TREE_EFFDT and T.TREE_NODE = A.DEPTID and T.TREE_NODE_NUM between S.TREE_NODE_NUM
and S.TREE_NODE_NUM_END
and not exists (select 'X' from PS_SCRTY_TBL_DEPT S1
where S.OPRID = S1.OPRID
and S.TREE_NODE_NUM <> S1.TREE_NODE_NUM
and T.TREE_NODE_NUM between S1.TREE_NODE_NUM
and S1.TREE_NODE_NUM_END
and S1.TREE_NODE_NUM between S.TREE_NODE_NUM
and S.TREE_NODE_NUM_END)
and B.EMPLID = A.EMPLID and B.EMPL_RCD_NBR = A.EMPL_RCD_NBR
and B.SINGLE_CHECK_USE IN ('C', 'N') and (B.REG_HRS<>0 or B.REG_HRLY_EARNS<>0)
and B.PAY_LINE_STATUS in ('C','F')
union
select A.EMPLID,S.OPRID,A.LASTUPDDTTM,A.EMPL_RCD_NBR,A.LAST_NAME,A.FIRST_NAME,A.DEPTID,B.COMPANY,B.PAYGROUP,B.PAY_END_DT,B.OFF_CYCLE,
B.ERNCD_OT_HRS,B.OT_HRS,0,B.OT_HRLY_EARNS
from PS_EMPLOYEES A,PS_PAY_EARNINGS B,PS_SCRTY_TBL_DEPT S,PSTREENODE T
where S.ACCESS_CD = 'Y' and T.SETID = ' ' and T.TREE_NAME = 'DEPT_SECURITY'
and T.EFFDT = S.TREE_EFFDT and T.TREE_NODE = A.DEPTID and T.TREE_NODE_NUM between S.TREE_NODE_NUM
and S.TREE_NODE_NUM_END
and not exists (select 'X' from PS_SCRTY_TBL_DEPT S1
where S.OPRID = S1.OPRID
and S.TREE_NODE_NUM <> S1.TREE_NODE_NUM
and T.TREE_NODE_NUM between S1.TREE_NODE_NUM
and S1.TREE_NODE_NUM_END
and S1.TREE_NODE_NUM between S.TREE_NODE_NUM
and S.TREE_NODE_NUM_END)
and B.EMPLID = A.EMPLID and B.EMPL_RCD_NBR = A.EMPL_RCD_NBR
and B.SINGLE_CHECK_USE IN ('C', 'N')
and (B.OT_HRS<>0 or B.OT_HRLY_EARNS<>0)
union
select
A.EMPLID,S.OPRID,A.LASTUPDDTTM,A.EMPL_RCD_NBR,A.LAST_NAME,A.FIRST_NAME,A.DEPTID,B.COMPANY,B.PAYGROUP,B.PAY_END_DT,B.OFF_CYCLE,
C.ERNCD,C.OTH_HRS,C.OTH_PAY,C.OTH_EARNS
from PS_EMPLOYEES A,PS_PAY_EARNINGS B,PS_PAY_OTH_EARNS C,PS_SCRTY_TBL_DEPT S,PSTREENODE T
where S.ACCESS_CD = 'Y' and T.SETID = ' ' and T.TREE_NAME = 'DEPT_SECURITY'
and T.EFFDT = S.TREE_EFFDT and T.TREE_NODE = A.DEPTID and T.TREE_NODE_NUM between S.TREE_NODE_NUM
and S.TREE_NODE_NUM_END
and not exists (select 'X' from PS_SCRTY_TBL_DEPT S1
where S.OPRID = S1.OPRID
and S.TREE_NODE_NUM <> S1.TREE_NODE_NUM
and T.TREE_NODE_NUM between S1.TREE_NODE_NUM
and S1.TREE_NODE_NUM_END
and S1.TREE_NODE_NUM between S.TREE_NODE_NUM
and S.TREE_NODE_NUM_END)
and B.EMPLID = A.EMPLID and B.EMPL_RCD_NBR = A.EMPL_RCD_NBR
and C.COMPANY = B.COMPANY and C.PAYGROUP = B.PAYGROUP
and C.PAY_END_DT = B.PAY_END_DT and C.OFF_CYCLE = B.OFF_CYCLE
and C.PAGE_NBR = B.PAGE_NBR and C.LINE_NBR = B.LINE_NBR
and C.ADDL_NBR = B.ADDL_NBR and B.SINGLE_CHECK_USE IN ('C', 'N')
and (C.OTH_HRS<>0 or C.OTH_PAY<>0 or C.OTH_EARNS<>0)
and B.PAY_LINE_STATUS in ('C','F')

Answer 1

UNION will eliminate the duplicates. If you are seeing rows that look like duplicates, there must be something different about them that you are missing.
If there is some set of rows that is common to all of the UNION queries and expensive to retrieve, you might try subquery factoring, e.g:
WITH common_set AS
    ( SELECT somecols
      FROM   sometab x, othertab y
      WHERE  y.key = x.key
      AND    etc )
SELECT a,b,c
FROM   common_set c, table1 t
WHERE  t.key = c.key
UNION
SELECT a,b,c
FROM   common_set c, table2 t
WHERE  t.key = c.key
UNION
SELECT a,b,c
FROM   common_set c, table2 t
WHERE  t.key = c.keyThe usual advice applies for tuning requests applies though:
<ul><li>{thread:id=863295}</li>
<li>{thread:id=501834}</li>
<li>Troubleshooting Bad Execution Plans</li>
<li>Writing Good SQL</li></ul>