上周五一哥们发了条SQL，让我看看，代码如下：

  1. SELECT COUNT(1)
  2.   FROM (select m.sheet_id
  3.           from cpm_main_sheet_history m, cpm_service_warn_config s
  4.          where m.sheet_type_id in
  5.                (select t.row_id
  6.                   from  tbl_class_trees   t
  7.                  start with t.row_id = s.sheet_type_id
  8.                 connect by t.parent_row_id = prior t.row_id)
  9.            and m.service_type  in
  10.                (select tt.row_id
  11.                   from  tbl_class_trees  tt
  12.                  start with tt.row_id = s.business_type_id
  13.                 connect by tt.parent_row_id = prior tt.row_id)
  14.            and m.accept_time >=
  15.                TO_CHAR(SYSDATE - time_interval / 24, 'yyyy-mm-dd hh24:mi:ss')
  16.            and s.row_id = 'AS170904165251'
  17.            and m.no_area in ('0000')) c__
  19. --执行计划
  20.  PLAN_TABLE_OUTPUT
  21. Plan hash value: 2710926849
  23. ----------------------------------------------------------------------------------------------------------------------
  24. | Id  | Operation                                  | Name                    | Rows  | Bytes | Cost (%CPU)| Time     |
  25. ----------------------------------------------------------------------------------------------------------------------
  26. |   0 | SELECT STATEMENT                           |                         |     1 |   143 | 96246   (1)| 00:19:15 |
  27. |*  1 |  FILTER                                    |                         |       |       |            |          |
  28. |   2 |   NESTED LOOPS                             |                         |  1681 |   234K|   746   (1)| 00:00:09 |
  29. |   3 |    TABLE ACCESS BY INDEX ROWID             | CPM_SERVICE_WARN_CONFIG |     1 |    56 |     1   (0)| 00:00:01 |
  30. |*  4 |     INDEX UNIQUE SCAN                      | PK_CONFIG_ROW_ID        |     1 |       |     0   (0)| 00:00:01 |
  31. |   5 |    TABLE ACCESS BY INDEX ROWID             | CPM_MAIN_SHEET_HISTORY  |  1681 |   142K|   745   (1)| 00:00:09 |
  32. |*  6 |     INDEX RANGE SCAN                       | IDX_CPM_NO_AREA_TIME1   |   449 |       |    62   (0)| 00:00:01 |
  33. |*  7 |   FILTER                                   |                         |       |       |            |          |
  34. |*  8 |    CONNECT BY NO FILTERING WITH SW (UNIQUE)|                         |       |       |            |          |
  35. |   9 |     TABLE ACCESS FULL                      | TBL_CLASS_TREES         |  9527 |   493K|   113   (0)| 00:00:02 |
  36. |* 10 |   FILTER                                   |                         |       |       |            |          |
  37. |* 11 |    CONNECT BY NO FILTERING WITH SW (UNIQUE)|                         |       |       |            |          |
  38. |  12 |     TABLE ACCESS FULL                      | TBL_CLASS_TREES         |  9527 |   493K|   113   (0)| 00:00:02 |
  39. ----------------------------------------------------------------------------------------------------------------------
  41. Predicate Information (identified by operation id):
  42. ---------------------------------------------------
  44.    1 - filter( EXISTS (SELECT 0 FROM "TBL_CLASS_TREES" "T" WHERE "T"."ROW_ID"=:B1 START WITH "T"."ROW_ID"=:B2
  45.               CONNECT BY "T"."PARENT_ROW_ID"=PRIOR "T"."ROW_ID") AND  EXISTS (SELECT 0 FROM "TBL_CLASS_TREES" "TT" WHERE
  46.               "TT"."ROW_ID"=:B3 START WITH "TT"."ROW_ID"=:B4 CONNECT BY "TT"."PARENT_ROW_ID"=PRIOR "TT"."ROW_ID"))
  47.    4 - access("S"."ROW_ID"='AS170904165251')
  48.    6 - access("M"."ACCEPT_TIME">=TO_CHAR(SYSDATE@!-"TIME_INTERVAL"/24,'yyyy-mm-dd hh24:mi:ss') AND
  49.               "M"."NO_AREA"='0000' AND "M"."ACCEPT_TIME" IS NOT NULL)
  50.        filter("M"."NO_AREA"='0000')
  51.    7 - filter("T"."ROW_ID"=:B1)
  52.    8 - access("T"."PARENT_ROW_ID"=PRIOR "T"."ROW_ID")
  53.        filter("T"."ROW_ID"=:B1)
  54.   10 - filter("TT"."ROW_ID"=:B1)
  55.   11 - access("TT"."PARENT_ROW_ID"=PRIOR "TT"."ROW_ID")
  56.        filter("TT"."ROW_ID"=:B1)

SQL优化前：

耗时：20s

count(1)返回: 147条数据

分析执行计划，执行计划中有filter关键字且有3个子级，这种sql是最容易引起性能问题的，所以第一时间是反应是sql有没有走索引，能不能改写。

尝试1：

建索引优化：

  1. 在TBL_CLASS_TREES表(row_id,parent_row_id)上建索引
  2. 执行计划：
  3. PLAN_TABLE_OUTPUT
  4. Plan hash value: 135779572
  6. ----------------------------------------------------------------------------------------------------------------------
  7. | Id  | Operation                                  | Name                    | Rows  | Bytes | Cost (%CPU)| Time     |
  8. ----------------------------------------------------------------------------------------------------------------------
  9. |   0 | SELECT STATEMENT                           |                         |     1 |   155 | 34147   (1)| 00:06:50 |
  10. |*  1 |  FILTER                                    |                         |       |       |            |          |
  11. |   2 |   NESTED LOOPS                             |                         |  3021 |   457K|   816   (1)| 00:00:10 |
  12. |   3 |    TABLE ACCESS BY INDEX ROWID             | CPM_SERVICE_WARN_CONFIG |     1 |    62 |     1   (0)| 00:00:01 |
  13. |*  4 |     INDEX UNIQUE SCAN                      | PK_CONFIG_ROW_ID        |     1 |       |     0   (0)| 00:00:01 |
  14. |   5 |    TABLE ACCESS BY INDEX ROWID             | CPM_MAIN_SHEET_HISTORY  |  3021 |   274K|   815   (1)| 00:00:10 |
  15. |*  6 |     INDEX RANGE SCAN                       | IDX_CPM_NO_AREA_TIME1   |   563 |       |   136   (0)| 00:00:02 |
  16. |*  7 |   FILTER                                   |                         |       |       |            |          |
  17. |*  8 |    CONNECT BY NO FILTERING WITH SW (UNIQUE)|                         |       |       |            |          |
  18. |   9 |     INDEX FAST FULL SCAN                   | IDX_ROW_ID              |  9527 |   493K|    20   (0)| 00:00:01 |
  19. |* 10 |   FILTER                                   |                         |       |       |            |          |
  20. |* 11 |    CONNECT BY NO FILTERING WITH SW (UNIQUE)|                         |       |       |            |          |
  21. |  12 |     INDEX FAST FULL SCAN                   | IDX_ROW_ID              |  9527 |   493K|    20   (0)| 00:00:01 |
  22. ----------------------------------------------------------------------------------------------------------------------
  24. Predicate Information (identified by operation id):
  25. ---------------------------------------------------
  27.    1 - filter( EXISTS (SELECT 0 FROM "TBL_CLASS_TREES" "T" WHERE "T"."ROW_ID"=:B1 START WITH "T"."ROW_ID"=:B2
  28.               CONNECT BY "T"."PARENT_ROW_ID"=PRIOR "T"."ROW_ID") AND  EXISTS (SELECT 0 FROM "TBL_CLASS_TREES" "TT" WHERE
  29.               "TT"."ROW_ID"=:B3 START WITH "TT"."ROW_ID"=:B4 CONNECT BY "TT"."PARENT_ROW_ID"=PRIOR "TT"."ROW_ID"))
  30.    4 - access("S"."ROW_ID"='AS170904165251')
  31.    6 - access("M"."ACCEPT_TIME">=TO_CHAR(SYSDATE@!-"TIME_INTERVAL"/24,'yyyy-mm-dd hh24:mi:ss') AND
  32.               "M"."NO_AREA"='0000' AND "M"."ACCEPT_TIME" IS NOT NULL)
  33.        filter("M"."NO_AREA"='0000')
  34.    7 - filter("T"."ROW_ID"=:B1)
  35.    8 - access("T"."PARENT_ROW_ID"=PRIOR "T"."ROW_ID")
  36.        filter("T"."ROW_ID"=:B1)
  37.   10 - filter("TT"."ROW_ID"=:B1)
  38.   11 - access("TT"."PARENT_ROW_ID"=PRIOR "TT"."ROW_ID")
  39.        filter("TT"."ROW_ID"=:B1)
  41. --效果还是一样慢，此优化失败。

尝试2：

利用with改写sql优化

  1. with t as (select /*+ materialize */ row_id,parent_row_id from tbl_class_trees)
  2. SELECT COUNT(1)
  3.   FROM (select m.sheet_id
  4.           from cpm_main_sheet_history m,cpm_service_warn_config s
  5.          where m.sheet_type_id in
  6.                (select t.row_id
  7.                   from  t
  8.                  start with t.row_id = s.sheet_type_id
  9.                 connect by t.parent_row_id = prior t.row_id)
  10.            and m.service_type  in
  11.                (select t.row_id
  12.                   from t
  13.                  start with t.row_id = s.business_type_id
  14.                 connect by t.parent_row_id = prior t.row_id)
  15.            and m.accept_time >=
  16.                TO_CHAR(SYSDATE - time_interval / 24, 'yyyy-mm-dd hh24:mi:ss')
  17.            and s.row_id = 'AS170904165251'
  18.            and m.no_area in ('0000')) c__
  20. --效果还是一样慢，此优化失败。

再次分析原SQL执行计划：

  1. id=8,id=11的执行计划关键词是：CONNECT BY NO FILTERING WITH SW (UNIQUE)。
  2. 这个为树形查询在11g中的新特性，尝试让sql不使用这个新特性。
  4. 于是使用以下hint：/*+ connect_by_filtering */  进行优化：
  6. SELECT COUNT(1)
  7.   FROM (select m.sheet_id
  8.           from cpm_main_sheet_history m, cpm_service_warn_config s
  9.          where m.sheet_type_id in
  10.                (select /*+ connect_by_filtering */ t.row_id
  11.                   from  tbl_class_trees   t
  12.                  start with t.row_id = s.sheet_type_id
  13.                 connect by t.parent_row_id = prior t.row_id)
  14.            and m.service_type  in
  15.                (select /*+ connect_by_filtering */ tt.row_id
  16.                   from  tbl_class_trees  tt
  17.                  start with tt.row_id = s.business_type_id
  18.                 connect by tt.parent_row_id = prior tt.row_id)
  19.            and m.accept_time >=
  20.                TO_CHAR(SYSDATE - time_interval / 24, 'yyyy-mm-dd hh24:mi:ss')
  21.            and s.row_id = 'AS170904165251'
  22.            and m.no_area in ('0000')) c__  
  24. PLAN_TABLE_OUTPUT
  25. Plan hash value: 2824841339
  27. ----------------------------------------------------------------------------------------------------------
  28. | Id  | Operation                      | Name                    | Rows  | Bytes | Cost (%CPU)| Time     |
  29. ----------------------------------------------------------------------------------------------------------
  30. |   0 | SELECT STATEMENT               |                         |     1 |   246 |   188K  (1)| 00:37:47 |
  31. |*  1 |  FILTER                        |                         |       |       |            |          |
  32. |   2 |   NESTED LOOPS                 |                         |  3021 |   725K|   816   (1)| 00:00:10 |
  33. |   3 |    TABLE ACCESS BY INDEX ROWID | CPM_SERVICE_WARN_CONFIG |     1 |   106 |     1   (0)| 00:00:01 |
  34. |*  4 |     INDEX UNIQUE SCAN          | PK_CONFIG_ROW_ID        |     1 |       |     0   (0)| 00:00:01 |
  35. |   5 |    TABLE ACCESS BY INDEX ROWID | CPM_MAIN_SHEET_HISTORY  |  3021 |   413K|   815   (1)| 00:00:10 |
  36. |*  6 |     INDEX RANGE SCAN           | IDX_CPM_NO_AREA_TIME1   |   563 |       |   136   (0)| 00:00:02 |
  37. |*  7 |   FILTER                       |                         |       |       |            |          |
  38. |*  8 |    CONNECT BY WITH FILTERING   |                         |       |       |            |          |
  39. |   9 |     TABLE ACCESS BY INDEX ROWID| TBL_CLASS_TREES         |     1 |   107 |     2   (0)| 00:00:01 |
  40. |* 10 |      INDEX UNIQUE SCAN         | PK_TBL_CLASS_TREESS     |     1 |       |     1   (0)| 00:00:01 |
  41. |* 11 |     HASH JOIN                  |                         |       |       |            |          |
  42. |  12 |      CONNECT BY PUMP           |                         |       |       |            |          |
  43. |  13 |      TABLE ACCESS FULL         | TBL_CLASS_TREES         |     6 |   318 |   113   (0)| 00:00:02 |
  44. |* 14 |   FILTER                       |                         |       |       |            |          |
  45. |* 15 |    CONNECT BY WITH FILTERING   |                         |       |       |            |          |
  46. |  16 |     TABLE ACCESS BY INDEX ROWID| TBL_CLASS_TREES         |     1 |   107 |     2   (0)| 00:00:01 |
  47. |* 17 |      INDEX UNIQUE SCAN         | PK_TBL_CLASS_TREESS     |     1 |       |     1   (0)| 00:00:01 |
  48. |* 18 |     HASH JOIN                  |                         |       |       |            |          |
  49. |  19 |      CONNECT BY PUMP           |                         |       |       |            |          |
  50. |  20 |      TABLE ACCESS FULL         | TBL_CLASS_TREES         |     6 |   318 |   113   (0)| 00:00:02 |
  51. ----------------------------------------------------------------------------------------------------------
  53. Predicate Information (identified by operation id):
  54. ---------------------------------------------------
  56.    1 - filter( EXISTS (SELECT /*+ CONNECT_BY_FILTERING */ 0 FROM "TBL_CLASS_TREES" "T" WHERE
  57.               "T"."ROW_ID"=:B1 START WITH "T"."ROW_ID"=:B2) AND  EXISTS (SELECT /*+ CONNECT_BY_FILTERING */ 0
  58.               FROM "TBL_CLASS_TREES" "TT" WHERE "TT"."ROW_ID"=:B3 START WITH "TT"."ROW_ID"=:B4))
  59.    4 - access("S"."ROW_ID"='AS170904165251')
  60.    6 - access("M"."ACCEPT_TIME">=TO_CHAR(SYSDATE@!-"TIME_INTERVAL"/24,'yyyy-mm-dd hh24:mi:ss')
  61.               AND "M"."NO_AREA"='0000' AND "M"."ACCEPT_TIME" IS NOT NULL)
  62.        filter("M"."NO_AREA"='0000')
  63.    7 - filter("T"."ROW_ID"=:B1)
  64.    8 - access("T"."PARENT_ROW_ID"=PRIOR "T"."ROW_ID")
  65.   10 - access("T"."ROW_ID"=:B1)
  66.   11 - access("T"."PARENT_ROW_ID"=PRIOR "T"."ROW_ID")
  67.   14 - filter("TT"."ROW_ID"=:B1)
  68.   15 - access("TT"."PARENT_ROW_ID"=PRIOR "TT"."ROW_ID")
  69.   17 - access("TT"."ROW_ID"=:B1)
  70.   18 - access("TT"."PARENT_ROW_ID"=PRIOR "TT"."ROW_ID")
  72. --优化后，SQL能在5s返回结果

树形查询SQL优化一例的更多相关文章

  1. 跨服务器查询sql语句样例

    若2个数据库在同一台机器上:insert into DataBase_A..Table1(col1,col2,col3----)select col11,col22,col33-- from Data ...

  2. 跨服务器查询sql语句样例（转）

    若2个数据库在同一台机器上: insert into DataBase_A..Table1(col1,col2,col3----) select col11,col22,col33-- from Da ...

  3. oracle 11g亿级复杂SQL优化一例（数量级性能提升）

    自从16年之后,因为工作原因,项目中就没有再使用oracle了,最近最近支持一个项目,又要开始负责这块事情了.最近在跑性能测试,配置全部调好之后,不少sql还存在性能低下的问题,主要涉及执行计划的不合 ...

  4. 查询SQL优化

    SQL优化的一般步骤 通过show status命令了解各种SQL的执行频率定位执行效率较低的SQL语句,重点select通过explain分析低效率的SQL确定问题并采取相应的优化措施 优化措施 s ...

  5. 反连接NOT EXISTS子查询中有or 谓词连接条件SQL优化一例

    背景 今天在日常数据库检查中,发现一SQL运行时间特别长,于是抓取出来,进行优化. 优化前: 耗时:503s 返回:0 SQL代码 SELECT * FROM MM_PAYABLEMONEY_TD P ...

  6. 1 min 数据查询 SQL 优化

    问题 前几天线上数据库 IOPS 飙升,一直居高不下,最近并没有升级.遂查看数据库正在执行的 SQL 语句,发现有个查询离线设备的语句极其缓慢. 探寻原因 SELECT o.* FROM ( SELE ...

  7. mysql联合查询sql优化

    我们在使用mysql数据库时,经常会使用到mysql的联合查询,联合查询分为内连接和外连接,内连接查询结果是联合的表都存在匹配才会有结果,外连接则根据驱动表是否存在匹配来生成结果集. 这里使用mysq ...

  8. oracle查询SQL优化相当重要

    如果表中的时间字段是索引,那么时间字段不要使用函数,函数会使索引失效. 例如: select * from mytable where trunc(createtime)=trunc(sysdate) ...

  9. Mysql 分页查询sql优化

    先查下数据表的总条数: SELECT COUNT(id) FROM ts_translation_send_address 执行分页界SQL 查看使用时间2.210s SELECT * FROM ts ...

随机推荐

  1. svn报错：privious operation has not finshed;run 'cleanup' if it was interrupted

    在更新svn的过程中,可能中途会取消,取消之后再次更新时可能提示,如下图: 下载sqlite3工具,进入此下载地址:https://www.sqlite.org/download.html 将sqli ...

  2. bzoj 1699: [Usaco2007 Jan]Balanced Lineup排队【st表||线段树】

    要求区间取min和max,可以用st表或线段树维护 st表 #include<iostream> #include<cstdio> using namespace std; c ...

  3. 51nod 1238 最小公倍数之和 V3 【欧拉函数+杜教筛】

    首先题目中给出的代码打错了,少了个等于号,应该是 G=0; for(i=1;i<=N;i++) for(j=1;j<=N;j++) { G = (G + lcm(i,j)) % 10000 ...

  4. 进击的Python【第十五章】：Web前端基础之DOM

    进击的Python[第十五章]:Web前端基础之DOM 简介:文档对象模型(Document Object Model,DOM)是一种用于HTML和XML文档的编程接口.它给文档提供了一种结构化的表示 ...

  5. 题解报告：hdu 1171 Big Event in HDU（多重背包）

    Problem Description Nowadays, we all know that Computer College is the biggest department in HDU. Bu ...

  6. C#中如何判断键盘按键和组合键

    好记性不如烂笔头子,现在记录下来,不一定会有很详尽的实例,只写最核心的部分. C# winform的窗体类有KeyPreview属性,可以接收窗体内控件的键盘事件注册.窗体和控件都有KeyDown,K ...

  7. 1270 数组的最大代价 dp

    http://www.51nod.com/onlineJudge/questionCode.html#!problemId=1270&judgeId=194704 一开始贪心,以为就两种情况, ...

  8. SpringMVC -- 必知必会

    SpringMVC基于模型--视图--控制器(Model-View-Controller,MVC)模式实现,属于SpringFrameWork的后续产品,已经融合在SpringWebFlow里面.它通 ...

  9. java之java.lang.UnsupportedClassVersionError：com/mysql/jdbc/Driver : Unsupported major.minor version 52.0

    问题解释:jdk版本和mysql驱动版本不兼容,比如:jdk1.7与mysql-connector-java-5.xxx兼容,但与mysql-connector-java-6.xxx及以上不兼容

  10. js内置对象总结

    在js里,一切皆为或者皆可以被用作对象.可通过new一个对象或者直接以字面量形式创建变量(如var i="aaa"),所有变量都有对象的性质. 注意:通过字面量创建的对象在调用属性 ...