时间:2021-07-01 10:21:17 帮助过:86人阅读
filter这个词总让人很费解,它下一级可以挂 一个子节点,二个子节点,三个子节点...。挂一个子节点意思过滤,如对全表进行扫描后
挂一个子节点意思过滤,如对全表进行扫描后,,按照条件过滤,丢弃不满足条件的数据。
挂二个子节点类似是nest loop。
挂三个子节点类似1和2做nest loop,结果集再与3做nest loop(这个是我推测的)。
SQL> set linesize 300
SQL> set timing on
SQL> set autotrace traceonly
SQL> SELECT count(DISTINCT BI.RISK_BASE_ID) count
2 FROM GG_RISK_BASE_INFO BI,
3 GG_TASK_FORM_CLASSTEMPLATE FC,
4 GG_TASK_FORM_TEMPLATEVERSION FT,
5 GG_TASK_FORM_RELATIONTEMPLAT FR,
6 GG_TASK_WORK_SHEET WS,
7 GG_TASK_FORM_EXETASKRECORD FE
8 WHERE FC.RISK_BASE_ID = BI.RISK_BASE_ID
9 and FT.TEMPLATE_FID = FC.TEMPLATE_ID
10 and FR.TEMPLATE_VERSION_ID = FT.TEMPLATE_VERSION_ID
11 and FR.OBJECT_ID = WS.WORK_SHEET_ID
12 and FE.REL_TEMP_ID = FR.REL_TEMP_ID
13 and FR.REL_OBJECT_TYPE = '1'
14 AND FE.TASK_START_TIME > BI.RISK_ASSESS_TIME
15 and (exists
16 (SELECT 1
17 FROM GG_TASK_FORM_RISK_CTL_LOG CL
18 where CL.TEMPLATE_VERSION_ID = FT.TEMPLATE_VERSION_ID) or exists
19 (SELECT 1
20 FROM GG_TASK_FORM_NEW_RISK_INFO RI
21 where RI.REL_TEMP_ID = FR.REL_TEMP_ID));
已用时间: 00: 04: 10.36
执行计划
数据规模:
select count(1) from GG_RISK_BASE_INFO; --2828
select count(1) from GG_TASK_FORM_CLASSTEMPLATE; --5600
select count(1) from GG_TASK_FORM_TEMPLATEVERSION; --7646
select count(1) from GG_TASK_FORM_EXETASKRECORD; --33064
select count(1) from GG_TASK_FORM_RELATIONTEMPLAT; --67319
select count(1) from GG_TASK_FORM_RISK_CTL_LOG;--5210
select count(1) from GG_TASK_FORM_NEW_RISK_INFO;--27831
CREATE INDEX IDX_PTNRI_REL_TEMP_ID ON GG_TASK_FORM_NEW_RISK_INFO(REL_TEMP_ID);
CREATE INDEX IDX_PTFRCL_REL_T_V_ID ON GG_TASK_FORM_RISK_CTL_LOG(TEMPLATE_VERSION_ID);