我有一个id表,它在不同的表中有一个parentid,这在应用层构建了一个文件夹结构。我需要获取特定“根文件夹”
中所有ID的列表select count(id)
from t1, t2
where t1.id=t2.id
connect by prior t1.id = t2.parentid
start with t1.id in (select id from t3 where name = 'Root Folder')
t1 = 4,102,065行
t2 = 48,965,392行
t3 是一个视图
t4 是存储t3中名称的基础表(仅在下面的解释计划中引用)
现在这会返回正确的结果,但需要3分41秒(计数为3,257,847)
有没有更好的方法来执行此类查询或是我唯一的选项索引?
这是解释计划,我已经替换了真实的表和索引名称以匹配上面的示例:
Plan
SELECT STATEMENT ALL_ROWSCost: 14 Bytes: 20 Cardinality: 1
17 SORT GROUP BY Bytes: 20 Cardinality: 1
16 CONNECT BY WITH FILTERING
10 NESTED LOOPS Cost: 5 Bytes: 153 Cardinality: 3
8 NESTED LOOPS Cost: 4 Bytes: 27 Cardinality: 1
6 VIEW VIEW SYS.VW_NSO_1 Cost: 2 Bytes: 10 Cardinality: 1
5 HASH UNIQUE Bytes: 57 Cardinality: 1
4 NESTED LOOPS Cost: 2 Bytes: 57 Cardinality: 1
2 TABLE ACCESS BY INDEX ROWID TABLE T2 Cost: 1 Bytes: 40 Cardinality: 1
1 INDEX RANGE SCAN INDEX T2_NAME Cost: 1 Cardinality: 1
3 INDEX UNIQUE SCAN INDEX (UNIQUE) T3_ID Cost: 1 Bytes: 17 Cardinality: 1
7 INDEX UNIQUE SCAN INDEX (UNIQUE) T1_ID Cost: 1 Bytes: 17 Cardinality: 1
9 INDEX RANGE SCAN INDEX T2_ID_PARENTID Cost: 1 Bytes: 72 Cardinality: 3
15 NESTED LOOPS Cost: 7 Bytes: 867 Cardinality: 17
13 NESTED LOOPS Cost: 6 Bytes: 1,292 Cardinality: 38
11 CONNECT BY PUMP
12 INDEX RANGE SCAN INDEX T2_PARENTID_ID Cost: 1 Bytes: 312 Cardinality: 13
14 INDEX UNIQUE SCAN INDEX (UNIQUE) T1_ID Cost: 1 Bytes: 17 Cardinality: 1
答案 0 :(得分:0)
您唯一的选择是在t1.id和t2.parentid上创建索引,并确保通过分析执行计划来使用这些索引,如果是这种情况则使用提示。还要注意为t1,t2及其索引提供新的性能统计。