Query taking long

Question

HI I have a query which is running longer .
I have checked every option of tuning this .I have attached the explain plan for the same . It seems it is doing a cartisian product .
SELECT   analytic_source_cd,
           SUM (CASE WHEN pricing_dt = '24jan2014' THEN cnt ELSE 0 END)
              AS Prev_Count,
           SUM (CASE WHEN pricing_dt = '27jan2014' THEN cnt ELSE 0 END)
              AS Current_Count
    FROM   (SELECT   af.analytic_source_cd,
                       af.pricing_dt,
                       COUNT (DISTINCT af.fi_instrument_id) cnt
                FROM   analytics_fact af,
                       fund f,
                       instrument_alternate_id iai,
                       (SELECT   pricing_dt, vendor_instrument_id, index_cd
                          FROM   fi_idx_benchmark_holdings
                         WHERE   pricing_dt IN
                                       ('24jan2014', '27jan2014')
                        UNION
                        SELECT   pricing_dt, vendor_instrument_id, index_cd
                          FROM   fi_idx_forward_holdings
                         WHERE   pricing_dt IN
                                       ('24jan2014', '27jan2014')) bh
               WHERE     
                        af.pricing_dt = bh.pricing_dt
                       AND f.official_index = bh.index_cd
                       AND af.fi_instrument_id = iai.fi_instrument_id
                       AND bh.vendor_instrument_id = iai.alternate_id
                       AND iai.alternate_id_type_code IN ('FMR_CUSIP', 'CUSIP')
                       and  af.pricing_dt IN ('24jan2014', '27jan2014')
                       AND f.official_index IS NOT NULL
                       AND af.oad IS NOT NULL
            GROUP BY   af.analytic_source_cd, af.pricing_dt
GROUP BY   analytic_source_cd
ORDER BY   1;
Please check the below .
Plan
SELECT STATEMENT  ALL_ROWSCost: 210,133  Bytes: 27  Cardinality: 1              
27 SORT GROUP BY  Cost: 210,133  Bytes: 27  Cardinality: 1             
  26 VIEW A519350. Cost: 210,133  Bytes: 27  Cardinality: 1            
   25 HASH GROUP BY  Cost: 210,133  Bytes: 26  Cardinality: 1           
    24 VIEW VIEW SYS.VM_NWVW_1 Cost: 210,133  Bytes: 26  Cardinality: 1          
     23 HASH GROUP BY  Cost: 210,133  Bytes: 87  Cardinality: 1         
      22 HASH JOIN  Cost: 210,132  Bytes: 87  Cardinality: 1        
       10 MERGE JOIN CARTESIAN  Cost: 130,054  Bytes: 63  Cardinality: 1       
        7 NESTED LOOPS  Cost: 129,831  Bytes: 61  Cardinality: 1      
         4 INLIST ITERATOR     
          3 PARTITION RANGE ITERATOR  Cost: 129,827  Bytes: 30  Cardinality: 1  Partition #: 10  Partitions accessed #KEY(INLIST)  
           2 TABLE ACCESS BY LOCAL INDEX ROWID TABLE FI_PORTFOLIO_DM.ANALYTICS_FACT Cost: 129,827  Bytes: 30  Cardinality: 1  Partition #: 10  Partitions accessed #KEY(INLIST) 
            1 INDEX RANGE SCAN INDEX (UNIQUE) FI_PORTFOLIO_DM.ANALYTICS_FACT_PK Cost: 667  Cardinality: 206,474  Partition #: 10  Partitions accessed #KEY(INLIST)
         6 PARTITION LIST INLIST  Cost: 4  Bytes: 31  Cardinality: 1  Partition #: 13  Partitions accessed #KEY(INLIST)   
          5 INDEX RANGE SCAN INDEX (UNIQUE) FI_REFERENCE.INSTRUMENT_ALTERNATE_ID_PPK Cost: 4  Bytes: 31  Cardinality: 1  Partition #: 13  Partitions accessed #KEY(INLIST)  
        9 BUFFER SORT  Cost: 130,050  Bytes: 1,642  Cardinality: 821      
         8 TABLE ACCESS FULL TABLE FI_REFERENCE.FUND Cost: 224  Bytes: 1,642  Cardinality: 821     
       21 VIEW A519350. Cost: 80,049  Bytes: 63,861,216  Cardinality: 2,660,884       
        20 SORT UNIQUE  Cost: 80,049  Bytes: 66,522,100  Cardinality: 2,660,884      
         19 UNION-ALL     
          14 INLIST ITERATOR    
           13 PARTITION RANGE ITERATOR  Cost: 24,599  Bytes: 25,284,850  Cardinality: 1,011,394  Partition #: 21  Partitions accessed #KEY(INLIST) 
            12 TABLE ACCESS BY LOCAL INDEX ROWID TABLE FI_BENCHMARK.FI_IDX_BENCHMARK_HOLDINGS Cost: 24,599  Bytes: 25,284,850  Cardinality: 1,011,394  Partition #: 21  Partitions accessed #KEY(INLIST)
             11 INDEX RANGE SCAN INDEX FI_BENCHMARK.FI_IDX_BENCHMARK_HOLDINGS_I2 Cost: 1,973  Cardinality: 1,011,394  Partition #: 21  Partitions accessed #KEY(INLIST)
          18 INLIST ITERATOR    
           17 PARTITION RANGE ITERATOR  Cost: 36,066  Bytes: 41,237,250  Cardinality: 1,649,490  Partition #: 25  Partitions accessed #KEY(INLIST) 
            16 TABLE ACCESS BY LOCAL INDEX ROWID TABLE FI_BENCHMARK.FI_IDX_FORWARD_HOLDINGS Cost: 36,066  Bytes: 41,237,250  Cardinality: 1,649,490  Partition #: 25  Partitions accessed #KEY(INLIST)
             15 INDEX RANGE SCAN INDEX FI_BENCHMARK.FI_IDX_FORWARD_HOLDINGS_I2 Cost: 3,499  Cardinality: 1,649,490  Partition #: 25  Partitions accessed #KEY(INLIST)
could you please help if i miss anything?

Answer 1

One nice best practice: do not hard code date columns, use TO_DATE function instead.
For performance issue, check the order of join tables. For ex, you need just af.pricing_dt IN ('24jan2014', '27jan2014') date range in af table, but you join all columns matching with bh table first. Then, you give date condition for af table again. Therefore, the intermediate processed rows will be higher.
Another nice best practice: You can use JOIN keywords, while joining tables. Writing all in where clause make the code complicated. Simplicity is not easy, but impressive.
Regards,
Dilek