Need help in executing sql

Question

Below is my sql which is taking forever...can any one tell whats wrong
SELECT COUNT(ACCOUNT_NBR) FROM STG_ACCOUNT WHERE ACCOUNT_NBR
NOT IN(SELECT OU_NUM FROM QA_ACCOUNT)
Below is plan for above query
Plan
SELECT STATEMENT CHOOSE               
     4 SORT AGGREGATE           
          3 FILTER      
               1 TABLE ACCESS FULL TABLE STG_ACCOUNT
               2 TABLE ACCESS FULL TABLE QA_ACCOUNT
Message was edited by:
user632733

Answer 1

As the execution plan shows, for each row from the STG_Account table full scan of QA_Account table is performed. It's VERY inefficient (each test tables has 50000 rows):
SQL> set timing on
SQL> set linesize 200
SQL> set autotrace on
SQL>
SQL> SELECT COUNT(ACCOUNT_NBR) FROM STG_ACCOUNT
  2  WHERE ACCOUNT_NBR NOT IN(SELECT OU_NUM FROM QA_ACCOUNT);
COUNT(ACCOUNT_NBR)
                 0
Elapsed: 00:02:27.31
Execution Plan
Plan hash value: 2866032720
| Id  | Operation           | Name        | Rows  | Bytes | Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT    |             |     1 |    13 | 32524   (4)| 00:06:31 |
|   1 |  SORT AGGREGATE     |             |     1 |    13 |            |          |
|*  2 |   FILTER            |             |       |       |            |          |
|   3 |    TABLE ACCESS FULL| STG_ACCOUNT | 52597 |   667K|    20   (5)| 00:00:01 |
|*  4 |    TABLE ACCESS FULL| QA_ACCOUNT  | 47771 |   606K|    20   (5)| 00:00:01 |
Predicate Information (identified by operation id):
   2 - filter( NOT EXISTS (SELECT /*+ */ 0 FROM "QA_ACCOUNT" "QA_ACCOUNT"
              WHERE LNNVL("OU_NUM"<>:B1)))
   4 - filter(LNNVL("OU_NUM"<>:B1))
Note
   - dynamic sampling used for this statement
Statistics
        478  recursive calls
          0  db block gets
    2072920  consistent gets
        169  physical reads
          0  redo size
        420  bytes sent via SQL*Net to client
        384  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
          8  sorts (memory)
          0  sorts (disk)
          1  rows processed
SQL> select value as "table scan rows gotten" from v$sesstat where statistic# = 249 and sid= [cur session sid] ;
table scan rows gotten
            1266674450Obviously, query should be rewritten, for example (db restarted):
SQL> set timing on
SQL> set linesize 200
SQL> set autotrace on
SQL> select count(*) from
  2  (
  3    select qa.ou_num
  4    from stg_account stg
  5      left outer join qa_account qa on stg.account_nbr = qa.ou_num
  6  )
  7  where ou_num is null
  8  ;
  COUNT(*)
         0
Elapsed: 00:00:00.29
Execution Plan
Plan hash value: 2130409646
| Id  | Operation             | Name        | Rows  | Bytes |TempSpc| Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT      |             |     1 |    26 |       |   164   (2)| 00:00:02 |
|   1 |  SORT AGGREGATE       |             |     1 |    26 |       |            |          |
|*  2 |   HASH JOIN RIGHT ANTI|             |  2311 | 60086 |  1232K|   164   (2)| 00:00:02 |
|   3 |    TABLE ACCESS FULL  | QA_ACCOUNT  | 50286 |   638K|       |    20   (5)| 00:00:01 |
|   4 |    TABLE ACCESS FULL  | STG_ACCOUNT | 52597 |   667K|       |    20   (5)| 00:00:01 |
Predicate Information (identified by operation id):
   2 - access("STG"."ACCOUNT_NBR"="QA"."OU_NUM")
Note
   - dynamic sampling used for this statement
Statistics
        476  recursive calls
          0  db block gets
        364  consistent gets
        167  physical reads
          0  redo size
        410  bytes sent via SQL*Net to client
        384  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
         10  sorts (memory)
          0  sorts (disk)
          1  rows processed
SQL> select value as "table scan rows gotten" from v$sesstat where statistic# = 249 and sid = [cur session sid];
table scan rows gotten
                264703Compare 'consistent gets' and 'table scan rows gotten' statistics of both queries.
Regards,
Dima