我有一个运行正常的复杂Oracle UNION查询。由于必须遍历太多记录，因此查询执行需要时间。我需要分析此查询并尽可能进行优化。
SELECT DISTINCT APP.APPLICATION_ID, APE.ENTITY_ID ENTITY_ID, APE.PARENT_ENTITY_ID, RP.ROLE_ID, RP.ACTION_ID, COALESCE(APE_MLD.ENTITY_NAME, APE.ENTITY_NAME)
ENTITY_NAME, RP.CURRENT_RECORD_STATUS_ID, COALESCE(ACT_MLD.ACTION_NAME, ACT.ACTION_NAME) ACTION_NAME, APE.ENTITY_PATH, APE.ADMIN_TYPE_ID, APE.IS_ADMIN_PAGE,
'RO.CR_NO', APE.ENTITY_CODE
FROM AUTH_APPLICATION_ENTITIES APE
INNER JOIN AUTH_APPLICATIONS APP ON APE.APPLICATION_ID = APP.APPLICATION_ID
LEFT JOIN AUTH_APPLICATION_ENTITIES_MLD APE_MLD ON APE.ENTITY_ID = APE_MLD.ENTITY_ID AND APE_MLD.LANGUAGE_ID = 1
LEFT JOIN AUTH_APPLICATION_ENTITIES PARENT_APE ON APE.PARENT_ENTITY_ID = PARENT_APE.ENTITY_ID
LEFT JOIN AUTH_APPLICATION_ENTITIES_MLD PARENT_APE_MLD ON PARENT_APE.ENTITY_ID = PARENT_APE_MLD.ENTITY_ID AND PARENT_APE_MLD.LANGUAGE_ID = 1
INNER JOIN AUTH_ROLES_PRIVILEGES RP ON APE.ENTITY_ID = RP.ENTITY_ID AND RP.CURRENT_RECORD_STATUS_ID NOT IN (2, 3, 6, 8, 9, 10)
INNER JOIN AUTH_ACTIONS ACT ON RP.ACTION_ID = ACT.ACTION_ID
LEFT JOIN AUTH_ACTIONS_MLD ACT_MLD ON ACT.ACTION_ID = ACT_MLD.ACTION_ID AND ACT_MLD.LANGUAGE_ID = 1
INNER JOIN AUTH_USERS_ROLES RO ON RP.ROLE_ID = RO.ROLE_ID AND RO.CURRENT_RECORD_STATUS_ID NOT IN (2, 3, 6, 8, 9, 10)
WHERE APE.CURRENT_RECORD_STATUS_ID ! = 2
AND RP.CURRENT_RECORD_STATUS_ID ! = 2
AND (APE.APPLICATION_ID = 715 OR 715 = -99)
AND RO.USER_ID = 1255162
AND (RO.ROLE_ID = -99 OR -99 = -99)
AND (APE.IS_ADMIN_PAGE = 0 OR USER_HAS_ADMIN_PRIVILEGES(APE.APPLICATION_ID, APE.ADMIN_TYPE_ID, 1255162) = 1)
UNION
SELECT DISTINCT APP.APPLICATION_ID, APE.ENTITY_ID ENTITY_ID, APE.PARENT_ENTITY_ID, RP.ROLE_ID, RP.ACTION_ID, COALESCE(APE_MLD.ENTITY_NAME, APE.ENTITY_NAME)
ENTITY_NAME, RP.CURRENT_RECORD_STATUS_ID, COALESCE(ACT_MLD.ACTION_NAME, ACT.ACTION_NAME) ACTION_NAME, APE.ENTITY_PATH, APE.ADMIN_TYPE_ID, APE.IS_ADMIN_PAGE,
'CR_NO' CR_NO, APE.ENTITY_CODE
FROM AUTH_APPLICATION_ENTITIES APE
INNER JOIN AUTH_APPLICATIONS APP ON APE.APPLICATION_ID = APP.APPLICATION_ID
LEFT JOIN AUTH_APPLICATION_ENTITIES_MLD APE_MLD ON APE.ENTITY_ID = APE_MLD.ENTITY_ID AND APE_MLD.LANGUAGE_ID = 1
LEFT JOIN AUTH_APPLICATION_ENTITIES PARENT_APE ON APE.PARENT_ENTITY_ID = PARENT_APE.ENTITY_ID
LEFT JOIN AUTH_APPLICATION_ENTITIES_MLD PARENT_APE_MLD ON PARENT_APE.ENTITY_ID = PARENT_APE_MLD.ENTITY_ID AND PARENT_APE_MLD.LANGUAGE_ID = 1
INNER JOIN AUTH_ROLES_PRIVILEGES RP ON APE.ENTITY_ID = RP.ENTITY_ID AND RP.CURRENT_RECORD_STATUS_ID NOT IN (2, 3, 6, 8, 9, 10)
INNER JOIN AUTH_ACTIONS ACT ON RP.ACTION_ID = ACT.ACTION_ID
LEFT JOIN AUTH_ACTIONS_MLD ACT_MLD ON ACT.ACTION_ID = ACT_MLD.ACTION_ID AND ACT_MLD.LANGUAGE_ID = 1
INNER JOIN AUTH_ROLE_TYPE_ROLES RTO ON RP.ROLE_ID = RTO.ROLE_ID AND RTO.CURRENT_RECORD_STATUS_ID NOT IN (2, 3, 6, 8, 9, 10)
WHERE APE.CURRENT_RECORD_STATUS_ID ! = 2
AND RP.CURRENT_RECORD_STATUS_ID ! = 2
AND (APE.APPLICATION_ID = 715 OR 715 = -99)
AND RTO.ROLE_TYPE_ID IN (SELECT ROLE_TYPE_ID FROM AUTH_USER_TYPES WHERE USER_ID = 1255162)
AND (RTO.ROLE_ID = -99 OR -99 = -99)
AND (APE.IS_ADMIN_PAGE = 0 OR USER_HAS_ADMIN_PRIVILEGES(APE.APPLICATION_ID, APE.ADMIN_TYPE_ID, 1255162) = 1);
我已经生成了解释计划，它提供了以下结果：
----------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
----------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 40 | 11232 | 1485 (63)| 00:00:18 |
| 1 | SORT UNIQUE | | 40 | 11232 | 1485 (63)| 00:00:18 |
| 2 | UNION-ALL | | | | | |
|* 3 | HASH JOIN OUTER | | 26 | 7228 | 631 (13)| 00:00:08 |
| 4 | NESTED LOOPS OUTER | | 22 | 5016 | 625 (13)| 00:00:08 |
| 5 | NESTED LOOPS OUTER | | 22 | 4862 | 625 (13)| 00:00:08 |
| 6 | NESTED LOOPS | | 22 | 4774 | 625 (13)| 00:00:08 |
|* 7 | HASH JOIN OUTER | | 22 | 4686 | 625 (13)| 00:00:08 |
|* 8 | HASH JOIN | | 22 | 4246 | 622 (13)| 00:00:08 |
|* 9 | HASH JOIN | | 22 | 3872 | 620 (13)| 00:00:08 |
|* 10 | TABLE ACCESS BY INDEX ROWID| AUTH_APPLICATION_ENTITIES | 40 | 5920 | 7 (0)| 00:00:01 |
|* 11 | INDEX RANGE SCAN | AUTH_AE_IDX_APP | 49 | | 1 (0)| 00:00:01 |
| 12 | NESTED LOOPS | | 597 | 16716 | 612 (13)| 00:00:08 |
|* 13 | TABLE ACCESS FULL | AUTH_USERS_ROLES | 5 | 65 | 607 (13)| 00:00:08 |
|* 14 | INDEX RANGE SCAN | AUTH_ROLES_PRIVILEGES_UK1 | 131 | 1965 | 1 (0)| 00:00:01 |
| 15 | TABLE ACCESS FULL | AUTH_ACTIONS | 11 | 187 | 2 (0)| 00:00:01 |
|* 16 | TABLE ACCESS FULL | AUTH_ACTIONS_MLD | 9 | 180 | 2 (0)| 00:00:01 |
|* 17 | INDEX UNIQUE SCAN | PK_SYS_MODULES | 1 | 4 | 0 (0)| 00:00:01 |
|* 18 | INDEX UNIQUE SCAN | PK_SYS_ENTITIES | 1 | 4 | 0 (0)| 00:00:01 |
|* 19 | INDEX UNIQUE SCAN | AUTH_APPLICATION_ENTITIES_UK1 | 1 | 7 | 0 (0)| 00:00:01 |
|* 20 | TABLE ACCESS FULL | AUTH_APPLICATION_ENTITIES_MLD | 1262 | 63100 | 5 (0)| 00:00:01 |
|* 21 | HASH JOIN OUTER | | 14 | 4004 | 852 (17)| 00:00:11 |
| 22 | NESTED LOOPS OUTER | | 12 | 2832 | 847 (17)| 00:00:11 |
| 23 | NESTED LOOPS OUTER | | 12 | 2748 | 847 (17)| 00:00:11 |
| 24 | NESTED LOOPS | | 12 | 2700 | 847 (17)| 00:00:11 |
|* 25 | HASH JOIN OUTER | | 12 | 2652 | 847 (17)| 00:00:11 |
|* 26 | HASH JOIN | | 12 | 2412 | 844 (17)| 00:00:11 |
|* 27 | HASH JOIN | | 12 | 2208 | 842 (17)| 00:00:11 |
|* 28 | TABLE ACCESS BY INDEX ROWID| AUTH_APPLICATION_ENTITIES | 40 | 5920 | 7 (0)| 00:00:01 |
|* 29 | INDEX RANGE SCAN | AUTH_AE_IDX_APP | 49 | | 1 (0)| 00:00:01 |
| 30 | NESTED LOOPS | | 308 | 11088 | 834 (17)| 00:00:11 |
|* 31 | HASH JOIN | | 3 | 63 | 831 (17)| 00:00:10 |
|* 32 | TABLE ACCESS FULL | AUTH_USER_TYPES | 4 | 40 | 829 (17)| 00:00:10 |
|* 33 | TABLE ACCESS FULL | AUTH_ROLE_TYPE_ROLES | 9 | 99 | 2 (0)| 00:00:01 |
|* 34 | INDEX RANGE SCAN | AUTH_ROLES_PRIVILEGES_UK1 | 89 | 1335 | 1 (0)| 00:00:01 |
| 35 | TABLE ACCESS FULL | AUTH_ACTIONS | 11 | 187 | 2 (0)| 00:00:01 |
|* 36 | TABLE ACCESS FULL | AUTH_ACTIONS_MLD | 9 | 180 | 2 (0)| 00:00:01 |
|* 37 | INDEX UNIQUE SCAN | PK_SYS_MODULES | 1 | 4 | 0 (0)| 00:00:01 |
|* 38 | INDEX UNIQUE SCAN | PK_SYS_ENTITIES | 1 | 4 | 0 (0)| 00:00:01 |
|* 39 | INDEX UNIQUE SCAN | AUTH_APPLICATION_ENTITIES_UK1 | 1 | 7 | 0 (0)| 00:00:01 |
|* 40 | TABLE ACCESS FULL | AUTH_APPLICATION_ENTITIES_MLD | 1262 | 63100 | 5 (0)| 00:00:01 |
----------------------------------------------------------------------------------------------------------------------
我需要通过阅读解释计划结果来优化上述查询。我只需要知道如何找到JOIN订单和慢速查询等。反正可以帮我吗？
答案 0 :(得分：1)
由于难以解决，我们不知道您的sql丢失了时间。
使用提示/ * + GATHER_PLAN_STATISTICS * /运行查询，或使用以下命令更改会话 alter session set statistics_level ='ALL'，然后将计划结果放在此处：
选择* FROM表（DBMS_XPLAN.DISPLAY_CURSOR（FORMAT =>'ALLSTATS LAST'））;
或者，如果您希望轻松阅读，请使用DBMS_SQL_MONITOR来检查查询