海量数据查询速度优化

Question

SELECT *
  FROM (SELECT CASE
                 WHEN V2.FIELD1 > 0 THEN
                  'B'
                 ELSE
                  'R'
               END VFIELD1,
               CASE
                 WHEN V2.FIELD2 > 0 THEN
                  'B'
                 ELSE
                  'R'
               END VFIELD2,
               V2.FIELD3 || '''' || V2.FIELD4 ||
               V2.FIELD5 AS CTMODEL,
               (SELECT rtrim(LIST(DISTINCT FIELD6 || '/'), '/')
                  FROM TABLE2
                 WHERE ID = V2.ID) FIELD6LIST,
               (SELECT rtrim(LIST(DISTINCT FIELD7 || '/'), '/')
                  FROM TABLE3
                 WHERE ID = V2.ID) FIELD7LIST,              
               (SELECT T4.STATE
                  FROM TABLE4 T4
                 WHERE T4.CNO = V2.CINNO) INSTATE, 
               (SELECT T4.STATE
                  FROM TABLE4 T4
                 WHERE T4.CNO = V2.COUTNO) OUTSTATE,                                                         
               (SELECT CASE
                         WHEN COUNT(*) > 0 THEN
                          'Y'
                         ELSE
                          'N'
                       END
                  FROM TABLE5 T5
                  LEFT JOIN TABLE6 PARTITION(PARTCUR) T6
                    ON T5.MID = T6.MID
                  LEFT JOIN TABLE7 T7
                    ON T7.JCODE = T6.JCODE
                 WHERE T5.ID = V2.ID
                   AND T6.JCODE = 'C1'
                   and rownum = 1) ISMID, 
               (SELECT CASE
                         WHEN COUNT(*) > 0 THEN
                          'Y'
                         ELSE
                          'N'
                       END
                  FROM TABLE5 T5
                  LEFT JOIN TABLE6 PARTITION(PARTCUR) T6
                    ON T5.MID = T6.MID
                  LEFT JOIN TABLE7 T7
                    ON T7.JCODE = T6.JCODE
                 WHERE T5.ID = V2.ID
                   AND T6.JCODE = 'C2'
                   AND ROWNUM = 1) ISTOP, 
               (SELECT CASE
                         WHEN COUNT(V1.VID) > 0 THEN
                          'Y'
                         ELSE
                          'N'
                       END
                  FROM VIEW1 V1
                 WHERE V1.ID = V2.ID
                   AND V1.TYPE = 'T') ISPREE
          FROM VIEW2 V2) TBVIEW2大概有100多万条记录，VIEW1大概30W条记录，其他表不多，在不带任何where条件的情况下，如何最大地提升此查询的速度？请高手指点分不多了，下次有分单独开贴给，谢谢了！

Answer 1

创建中间过程临时表，将该大SQL拆分成几个小SQL，通过IO来换取CPU，提升性能

Answer 2

两种方法：
1、将view1 和 view2 的数据放入中间临时表，然后建立相关索引进行关联查询。
2、增加/* parallel(table_name,8) */ 这样的多线程操作提示
3、直接修改session目前的session多线程操作，然后再查询。
   alter session enable parallel query;

Answer 3

把以下的每一小段都分別寫成函數,然後再調整函數的方式來取得想要的值.(SELECT rtrim(LIST(DISTINCT FIELD6 || '/'), '/')
                  FROM TABLE2
                 WHERE ID = V2.ID) FIELD6LIST,
               (SELECT rtrim(LIST(DISTINCT FIELD7 || '/'), '/')
                  FROM TABLE3
                 WHERE ID = V2.ID) FIELD7LIST,              
               (SELECT T4.STATE
                  FROM TABLE4 T4
                 WHERE T4.CNO = V2.CINNO) INSTATE, 
               (SELECT T4.STATE
                  FROM TABLE4 T4
                 WHERE T4.CNO = V2.COUTNO) OUTSTATE,                                                         
               (SELECT CASE
                         WHEN COUNT(*) > 0 THEN
                          'Y'
                         ELSE
                          'N'
                       END
                  FROM TABLE5 T5
                  LEFT JOIN TABLE6 PARTITION(PARTCUR) T6
                    ON T5.MID = T6.MID
                  LEFT JOIN TABLE7 T7
                    ON T7.JCODE = T6.JCODE
                 WHERE T5.ID = V2.ID
                   AND T6.JCODE = 'C1'
                   and rownum = 1) ISMID, 
               (SELECT CASE
                         WHEN COUNT(*) > 0 THEN
                          'Y'
                         ELSE
                          'N'
                       END
                  FROM TABLE5 T5
                  LEFT JOIN TABLE6 PARTITION(PARTCUR) T6
                    ON T5.MID = T6.MID
                  LEFT JOIN TABLE7 T7
                    ON T7.JCODE = T6.JCODE
                 WHERE T5.ID = V2.ID
                   AND T6.JCODE = 'C2'
                   AND ROWNUM = 1) ISTOP, 
               (SELECT CASE
                         WHEN COUNT(V1.VID) > 0 THEN
                          'Y'
                         ELSE
                          'N'
                       END
                  FROM VIEW1 V1
                 WHERE V1.ID = V2.ID
                   AND V1.TYPE = 'T') ISPREE

Answer 4

根据以上的语句，可做以下的查询优化（数据量不是很大）：
（1）将view1 和view2建成物化视图，增量刷新
（2）建立索引，通过查询执行计划，确定所建立的索引能用上。提示：在sql中尽量少用函数，函数在执行计划中无法体现。