Not in and Not Exists

Question

Hi everyone,
I need some explanation on why the index is not being used in one of my queries.
Here are the queries:
SELECT COUNT (1)
FROM (SELECT hdr.ALLOC_HDR_SEQ_NBR hdrnbr
FROM sm_vda_allocation_header hdr
INNER JOIN
sm_vda_type vdatype
ON hdr.vda_typ_cd = vdatype.vda_typ_cd
WHERE
vdatype.SAP_POST_TYP_CD = 'JE'
AND vdatype.vda_alloc_flg = 'Y'
AND hdr.vda_liab_flg IS NULL
AND hdr.sap_extrct_dt IS NULL
AND hdr.ALLOC_HDR_SEQ_NBR NOT IN
(SELECT alloc_hdr_seq_nbr
FROM sm_vda_allocation_detail));
return 0 rows.
explain plan::
Plan
SELECT STATEMENT CHOOSECost: 671 Bytes: 26 Cardinality: 1                          
     8 SORT AGGREGATE Bytes: 26 Cardinality: 1                     
          7 HASH JOIN Cost: 671 Bytes: 293,254 Cardinality: 11,279                
               1 TABLE ACCESS FULL SM_APPL.SM_VDA_TYPE Cost: 5 Bytes: 56 Cardinality: 7           
               6 MERGE JOIN ANTI Cost: 665 Bytes: 281,106 Cardinality: 15,617           
                    3 TABLE ACCESS BY INDEX ROWID SM_APPL.SM_VDA_ALLOCATION_HEADER Cost: 496 Bytes: 241,860 Cardinality: 20,155      
                         2 INDEX FULL SCAN UNIQUE SM_APPL.SM_VDA_ALLOCATION_HEADER_PK Cost: 438 Cardinality: 20,450
                    5 SORT UNIQUE Cost: 169 Bytes: 417,480 Cardinality: 69,580      
                         4 INDEX FAST FULL SCAN NON-UNIQUE SM_APPL.SM_VDA_ALLOCATION_DETAIL_FK_I Cost: 86 Bytes: 417,480 Cardinality: 69,580
2nd query::
SELECT COUNT (1)
FROM (SELECT hdr.ALLOC_HDR_SEQ_NBR hdrnbr
FROM sm_vda_allocation_header hdr
INNER JOIN
sm_vda_type vdatype
ON hdr.vda_typ_cd = vdatype.vda_typ_cd
WHERE
vdatype.SAP_POST_TYP_CD = 'JE'
AND vdatype.vda_alloc_flg = 'Y'
AND hdr.vda_liab_flg IS NULL
AND hdr.sap_extrct_dt IS NULL
AND NOT exists
(SELECT 1
FROM sm_vda_allocation_detail b where
hdr.ALLOC_HDR_SEQ_NBR = b.ALLOC_HDR_SEQ_NBR));
explain plan:::
Plan
SELECT STATEMENT CHOOSECost: 831 Bytes: 20 Cardinality: 1                     
     6 SORT AGGREGATE Bytes: 20 Cardinality: 1                
          5 FILTER           
               3 HASH JOIN Cost: 828 Bytes: 14,560 Cardinality: 728      
                    1 TABLE ACCESS FULL SM_APPL.SM_VDA_TYPE Cost: 5 Bytes: 56 Cardinality: 7
                    2 TABLE ACCESS FULL SM_APPL.SM_VDA_ALLOCATION_HEADER Cost: 822 Bytes: 12,096 Cardinality: 1,008
               4 INDEX RANGE SCAN NON-UNIQUE SM_APPL.SM_VDA_ALLOCATION_DETAIL_FK_I Cost: 3 Bytes: 90 Cardinality: 15      
In the 2nd query with NOT EXISTS, I see FULL table scan for SM_APPL.SM_VDA_ALLOCATION_HEADER
In the first query, the index is used in the same place:
2 INDEX FULL SCAN UNIQUE SM_APPL.SM_VDA_ALLOCATION_HEADER_PK Cost: 438 Cardinality: 20,450
ALLOC_HDR_SEQ_NBR is a primary key column.
Actually NOT IN and NOT EXISTS both used index for the second inner table (SM_APPL.SM_VDA_ALLOCATION_DETAIL).
But NOT exists is not using the index for the first table but not in is using.
how come.

Answer 1

872605 wrote:
Hi everyone,
I need some explanation on why the index is not being used in one of my queries.
Here are the queries:
SELECT COUNT (1)
FROM (SELECT hdr.ALLOC_HDR_SEQ_NBR hdrnbr
FROM sm_vda_allocation_header hdr
INNER JOIN
sm_vda_type vdatype
ON hdr.vda_typ_cd = vdatype.vda_typ_cd
WHERE
vdatype.SAP_POST_TYP_CD = 'JE'
AND vdatype.vda_alloc_flg = 'Y'
AND hdr.vda_liab_flg IS NULL
AND hdr.sap_extrct_dt IS NULL
AND hdr.ALLOC_HDR_SEQ_NBR NOT IN
(SELECT alloc_hdr_seq_nbr
FROM sm_vda_allocation_detail));
return 0 rows.
explain plan::
Plan
SELECT STATEMENT CHOOSECost: 671 Bytes: 26 Cardinality: 1                          
     8 SORT AGGREGATE Bytes: 26 Cardinality: 1                     
          7 HASH JOIN Cost: 671 Bytes: 293,254 Cardinality: 11,279                
               1 TABLE ACCESS FULL SM_APPL.SM_VDA_TYPE Cost: 5 Bytes: 56 Cardinality: 7           
               6 MERGE JOIN ANTI Cost: 665 Bytes: 281,106 Cardinality: 15,617           
                    3 TABLE ACCESS BY INDEX ROWID SM_APPL.SM_VDA_ALLOCATION_HEADER Cost: 496 Bytes: 241,860 Cardinality: 20,155      
                         2 INDEX FULL SCAN UNIQUE SM_APPL.SM_VDA_ALLOCATION_HEADER_PK Cost: 438 Cardinality: 20,450
                    5 SORT UNIQUE Cost: 169 Bytes: 417,480 Cardinality: 69,580      
                         4 INDEX FAST FULL SCAN NON-UNIQUE SM_APPL.SM_VDA_ALLOCATION_DETAIL_FK_I Cost: 86 Bytes: 417,480 Cardinality: 69,580
2nd query::
SELECT COUNT (1)
FROM (SELECT hdr.ALLOC_HDR_SEQ_NBR hdrnbr
FROM sm_vda_allocation_header hdr
INNER JOIN
sm_vda_type vdatype
ON hdr.vda_typ_cd = vdatype.vda_typ_cd
WHERE
vdatype.SAP_POST_TYP_CD = 'JE'
AND vdatype.vda_alloc_flg = 'Y'
AND hdr.vda_liab_flg IS NULL
AND hdr.sap_extrct_dt IS NULL
AND NOT exists
(SELECT 1
FROM sm_vda_allocation_detail b where
hdr.ALLOC_HDR_SEQ_NBR = b.ALLOC_HDR_SEQ_NBR));
explain plan:::
Plan
SELECT STATEMENT CHOOSECost: 831 Bytes: 20 Cardinality: 1                     
     6 SORT AGGREGATE Bytes: 20 Cardinality: 1                
          5 FILTER           
               3 HASH JOIN Cost: 828 Bytes: 14,560 Cardinality: 728      
                    1 TABLE ACCESS FULL SM_APPL.SM_VDA_TYPE Cost: 5 Bytes: 56 Cardinality: 7
                    2 TABLE ACCESS FULL SM_APPL.SM_VDA_ALLOCATION_HEADER Cost: 822 Bytes: 12,096 Cardinality: 1,008
               4 INDEX RANGE SCAN NON-UNIQUE SM_APPL.SM_VDA_ALLOCATION_DETAIL_FK_I Cost: 3 Bytes: 90 Cardinality: 15      
In the 2nd query with NOT EXISTS, I see FULL table scan for SM_APPL.SM_VDA_ALLOCATION_HEADER
In the first query, the index is used in the same place:
2 INDEX FULL SCAN UNIQUE SM_APPL.SM_VDA_ALLOCATION_HEADER_PK Cost: 438 Cardinality: 20,450
ALLOC_HDR_SEQ_NBR is a primary key column.
Actually NOT IN and NOT EXISTS both used index for the second inner table (SM_APPL.SM_VDA_ALLOCATION_DETAIL).
But NOT exists is not using the index for the first table but not in is using.
how come.WHY MY INDEX IS NOT BEING USED
http://communities.bmc.com/communities/docs/DOC-10031
http://searchoracle.techtarget.com/tip/Why-isn-t-my-index-getting-used
http://www.orafaq.com/tuningguide/not%20using%20index.html