Oracle中如何優化connect by語句

Oracle中如何優化connect by語句，針對這個問題，這篇文章詳細介紹了相對應的分析和解答，希望可以幫助更多想解決這個問題的小伙伴找到更簡單易行的方法。

執行SQL:

SELECT A.CI, A.ENBAJ02 AS CELL_NAME   FROM TDL_CM_CELL A, T_ORG_CELL_SCOPE S  WHERE S.REGION_NAME = A.REGION_NAME   AND S.CITY_NAME = A.CITY_NAME   AND (S.ORG_ID) IN (SELECT ID                         FROM T_ORG O                       START WITH ID = 101021003 --1010210                         --START WITH ID=1                       CONNECT BY PARENT_ID = PRIOR ID)

實際使用的執行計劃:

而不會采用自適應計劃(adaptive plan)：

 Plan Hash Value  : 2596385940   ------------------------------------------------------------------------------------------------------------------- | Id  | Operation                                       | Name                  | Rows | Bytes  | Cost | Time     | ------------------------------------------------------------------------------------------------------------------- |   0 | SELECT STATEMENT                                |                       | 2622 | 228114 |  227 | 00:00:01 | |   1 |   NESTED LOOPS                                  |                       | 2622 | 228114 |  227 | 00:00:01 | |   2 |    NESTED LOOPS                                 |                       | 2622 | 228114 |  227 | 00:00:01 | | * 3 |     HASH JOIN                                   |                       |    1 |     31 |    7 | 00:00:01 | |   4 |      VIEW                                       | VW_NSO_1              |    1 |     13 |    4 | 00:00:01 | |   5 |       HASH UNIQUE                               |                       |    1 |     20 |    4 | 00:00:01 | | * 6 |        CONNECT BY NO FILTERING WITH SW (UNIQUE) |                       |      |        |      |          | |   7 |         TABLE ACCESS FULL                       | T_ORG                 |   75 |    825 |    3 | 00:00:01 | |   8 |      TABLE ACCESS FULL                          | T_ORG_CELL_SCOPE      |   85 |   1530 |    3 | 00:00:01 | | * 9 |     INDEX RANGE SCAN                            | IDX_TDL_CM_CELL_SCOPE |  257 |        |    8 | 00:00:01 | |  10 |    TABLE ACCESS BY INDEX ROWID                  | TDL_CM_CELL           | 2313 | 129528 |  220 | 00:00:01 | -------------------------------------------------------------------------------------------------------------------  Predicate Information (identified by operation id): ------------------------------------------ * 3 - access("S"."ORG_ID"="ID") * 6 - access("PARENT_ID"=PRIOR "ID") * 6 - filter("ID"=101021003) * 9 - access("S"."REGION_NAME"="A"."REGION_NAME" AND "S"."CITY_NAME"="A"."CITY_NAME")   Notes ----- - This is an adaptive plan

原因在于，oracle無法知道connect by之后的數量，所以只能認為是很大的量

--

有一種方式就是，就是使用提示來解決：

SELECT /*+ no_merge(x) use_nl(a x) */   A.CI, A.ENBAJ02 AS CELL_NAME    FROM TDL_CM_CELL A,         (select s.city_name, s.region_name            from T_ORG_CELL_SCOPE S           WHERE (S.ORG_ID) IN                 (SELECT ID                    FROM T_ORG O                   START WITH ID = 101021003 --1010210                    --START WITH ID=1                  CONNECT BY PARENT_ID = PRIOR ID)                    ) x   where x.REGION_NAME = A.REGION_NAME     AND x.CITY_NAME = A.CITY_NAME

這樣計劃就是：

Plan Hash Value  : 37846894   --------------------------------------------------------------------------------------------------------------------- | Id   | Operation                                        | Name                  | Rows | Bytes  | Cost | Time     | --------------------------------------------------------------------------------------------------------------------- |    0 | SELECT STATEMENT                                 |                       | 2313 | 277560 |  227 | 00:00:01 | |    1 |   NESTED LOOPS                                   |                       | 2313 | 277560 |  227 | 00:00:01 | |    2 |    NESTED LOOPS                                  |                       | 2313 | 277560 |  227 | 00:00:01 | |    3 |     VIEW                                         |                       |    1 |     64 |    7 | 00:00:01 | |  * 4 |      HASH JOIN                                   |                       |    1 |     31 |    7 | 00:00:01 | |    5 |       VIEW                                       | VW_NSO_1              |    1 |     13 |    4 | 00:00:01 | |    6 |        HASH UNIQUE                               |                       |    1 |     20 |    4 | 00:00:01 | |  * 7 |         CONNECT BY NO FILTERING WITH SW (UNIQUE) |                       |      |        |      |          | |    8 |          TABLE ACCESS FULL                       | T_ORG                 |   75 |    825 |    3 | 00:00:01 | |    9 |       TABLE ACCESS FULL                          | T_ORG_CELL_SCOPE      |   85 |   1530 |    3 | 00:00:01 | | * 10 |     INDEX RANGE SCAN                             | IDX_TDL_CM_CELL_SCOPE |  257 |        |    8 | 00:00:01 | |   11 |    TABLE ACCESS BY INDEX ROWID                   | TDL_CM_CELL           | 2313 | 129528 |  220 | 00:00:01 | ---------------------------------------------------------------------------------------------------------------------  Predicate Information (identified by operation id): ------------------------------------------ * 4 - access("S"."ORG_ID"="ID") * 7 - access("PARENT_ID"=PRIOR "ID") * 7 - filter("ID"=101021003) * 10 - access("X"."REGION_NAME"="A"."REGION_NAME" AND "X"."CITY_NAME"="A"."CITY_NAME")

關于Oracle中如何優化connect by語句問題的解答就分享到這里了，希望以上內容可以對大家有一定的幫助，如果你還有很多疑惑沒有解開，可以關注億速云行業資訊頻道了解更多相關知識。