2. 程式人生 > 實用技巧 >Oracle使用SPM優化指定SQL

Oracle使用SPM優化指定SQL

阿新 • • 發佈：2020-09-01

Oracle使用SPM優化指定SQL

環境構造

16:44:23 SYS@zkm(1)> drop table scott.zkm purge;

Table dropped.

Elapsed: 00:00:00.31
16:44:30 SYS@zkm(1)> create table scott.zkm as select * from dba_objects;

Table created.

Elapsed: 00:00:00.39
16:44:35 SYS@zkm(1)> create index scott.idx_object_id on scott.zkm(object_id) online;


 
Index created.

Elapsed: 00:00:00.21

目標SQL

select /*+ full(a) */ * from scott.zkm a where object_id=1000;

由於hint的強制關係，該SQL會執行全表掃描，如下：

16:44:42 SYS@zkm(1)> set autotrace traceonly 
16:46:53 SYS@zkm(1)> set line 500
16:47:02 SYS@zkm(1)> select /*+ full(a) */ * from scott.zkm a where object_id=1000;

Elapsed: 
 
00:00:00.05

Execution Plan
----------------------------------------------------------
Plan hash value: 1571665327

--------------------------------------------------------------------------
| Id  | Operation         | Name | Rows  | Bytes | Cost (%CPU)| Time     |
--------------------------------------------------------------------------
 

|   0 | SELECT STATEMENT  |      |     1 |   207 |   366   (1)| 00:00:05 |
|*  1 |  TABLE ACCESS FULL| ZKM  |     1 |   207 |   366   (1)| 00:00:05 |
--------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   1 - filter("OBJECT_ID"=1000)

Note
-----
   - dynamic sampling used for this statement (level=2)


Statistics
----------------------------------------------------------
        122  recursive calls
          0  db block gets
       1503  consistent gets
       1603  physical reads
          0  redo size
       1628  bytes sent via SQL*Net to client
        523  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
         13  sorts (memory)
          0  sorts (disk)
          1  rows processed

新增基線，

col SQL_HANDLE for a25
col plan_name for a35
select signature,sql_handle,plan_name,origin,enabled,accepted,autopurge from dba_sql_plan_baselines;
模板複製 

17:22:12 SYS@zkm(27)> col SQL_HANDLE for a25
17:22:21 SYS@zkm(27)> col plan_name for a35
17:22:26 SYS@zkm(27)> select signature,sql_handle,plan_name,origin,enabled,accepted,autopurge from dba_sql_plan_baselines;

 SIGNATURE SQL_HANDLE                PLAN_NAME                           ORIGIN                                     ENABLED   ACCEPTED  AUTOPURGE
---------- ------------------------- ----------------------------------- ------------------------------------------ --------- --------- ---------
8.8386E+18 SQL_7aa929320eda7c96      SQL_PLAN_7pa99687dnz4q45324a8c      MANUAL-LOAD                                YES       YES       YES

Elapsed: 00:00:00.00

可以檢視SQL_HANDLE='SQL_7aa929320eda7c96'且PLAN_NAME='SQL_PLAN_7pa99687dnz4q45324a8c'對應的執行計劃，

17:22:30 SYS@zkm(27)> select * from table(dbms_xplan.display_sql_plan_baseline(sql_handle=>'SQL_7aa929320eda7c96',plan_name=>'SQL_PLAN_7pa99687dnz4q45324a8c'));

PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

--------------------------------------------------------------------------------
SQL handle: SQL_7aa929320eda7c96
SQL text: select /*+ full(a) */ * from scott.zkm a where object_id=1000
--------------------------------------------------------------------------------

--------------------------------------------------------------------------------
Plan name: SQL_PLAN_7pa99687dnz4q45324a8c         Plan id: 1160923788
Enabled: YES     Fixed: NO      Accepted: YES     Origin: MANUAL-LOAD
--------------------------------------------------------------------------------

Plan hash value: 1571665327

--------------------------------------------------------------------------
| Id  | Operation         | Name | Rows  | Bytes | Cost (%CPU)| Time     |
--------------------------------------------------------------------------
|   0 | SELECT STATEMENT  |      |  1097 |   221K|   366   (1)| 00:00:05 |
|*  1 |  TABLE ACCESS FULL| ZKM  |  1097 |   221K|   366   (1)| 00:00:05 |
--------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   1 - filter("OBJECT_ID"=1000)

24 rows selected.

Elapsed: 00:00:00.07

若是庫快取中存在優秀的優秀的執行計劃，那麼直接找出plan hash value新增到SQL作為基線之一，然後刪除較差的sql的基線。

若是沒有優秀的執行計劃，可以手工使用hint產生你想要的執行計劃後，新增為基線，下邊演示這種。

新增hint強制走索引，

17:14:03 SYS@zkm(1)> select /*+ index(a idx_object_id) */ * from scott.zkm a where object_id=1000;

Elapsed: 00:00:00.00

Execution Plan
----------------------------------------------------------
Plan hash value: 3532417104

---------------------------------------------------------------------------------------------
| Id  | Operation                   | Name          | Rows  | Bytes | Cost (%CPU)| Time     |
---------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |               |     1 |   207 |     2   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| ZKM           |     1 |   207 |     2   (0)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN          | IDX_OBJECT_ID |     1 |       |     1   (0)| 00:00:01 |
---------------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   2 - access("OBJECT_ID"=1000)

Note
-----
   - dynamic sampling used for this statement (level=2)


Statistics
----------------------------------------------------------
         27  recursive calls
          0  db block gets
         91  consistent gets
          1  physical reads
          0  redo size
       1631  bytes sent via SQL*Net to client
        523  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
          4  sorts (memory)
          0  sorts (disk)
          1  rows processed

先查詢sql_id值，

col sql_id for a15
set line 500
col sql_text for a100
select sql_id,sql_text,PLAN_HASH_VALUE from v$sql where sql_text like 'select%from scott.zkm a where object_id=1000';
模板複製 

17:44:56 SYS@zkm(27)> col sql_id for a15
17:45:03 SYS@zkm(27)> set line 500
17:45:07 SYS@zkm(27)> col sql_text for a100
17:45:10 SYS@zkm(27)> select sql_id,sql_text,PLAN_HASH_VALUE from v$sql where sql_text like 'select%from scott.zkm a where object_id=1000';

SQL_ID          SQL_TEXT                                                                                             PLAN_HASH_VALUE
--------------- ---------------------------------------------------------------------------------------------------- ---------------
9ajk015s54vpv   select /*+ full(a) */ * from scott.zkm a where object_id=1000                                             1571665327
7akf5qzu2f7wn   select /*+ index(a idx_object_id) */ * from scott.zkm a where object_id=1000                              3532417104

Elapsed: 00:00:00.00

將上邊執行路徑走了索引的執行計劃新增至

select /*+ full(a) */ * from scott.zkm a where object_id=1000;對應基線

中。

DECLARE
 k1 pls_integer;
begin
k1 := DBMS_SPM.LOAD_PLANS_FROM_CURSOR_CACHE (
sql_id=>'7akf5qzu2f7wn',
plan_hash_value=>3532417104,sql_handle=>'SQL_7aa929320eda7c96'
);
end;
/
模板複製 
17:38:42 SYS@zkm(27)> DECLARE
17:39:00   2   k1 pls_integer;
17:39:00   3  begin
17:39:00   4  k1 := DBMS_SPM.LOAD_PLANS_FROM_CURSOR_CACHE (
17:39:00   5  sql_id=>'7akf5qzu2f7wn',
17:39:00   6  plan_hash_value=>3532417104,sql_handle=>'SQL_7aa929320eda7c96'
17:39:00   7  );
17:39:00   8  end;
17:39:00   9  /

PL/SQL procedure successfully completed.

Elapsed: 00:00:00.01

17:47:46 SYS@zkm(1)> col SQL_HANDLE for a25
17:47:47 SYS@zkm(1)> col plan_name for a35
17:47:47 SYS@zkm(1)> select signature,sql_handle,plan_name,origin,enabled,accepted,autopurge from dba_sql_plan_baselines;

 SIGNATURE SQL_HANDLE                PLAN_NAME                           ORIGIN                                     ENABLED   ACCEPTED  AUTOPURGE
---------- ------------------------- ----------------------------------- ------------------------------------------ --------- --------- ---------
8.8386E+18 SQL_7aa929320eda7c96      SQL_PLAN_7pa99687dnz4q45324a8c      MANUAL-LOAD                                YES       YES       YES
8.8386E+18 SQL_7aa929320eda7c96      SQL_PLAN_7pa99687dnz4q54f02d62      MANUAL-LOAD                                YES       YES       YES

對應執行計劃，

17:47:47 SYS@zkm(1)> select * from table(dbms_xplan.display_sql_plan_baseline(sql_handle=>'SQL_7aa929320eda7c96',plan_name=>'SQL_PLAN_7pa99687dnz4q54f02d62'));

PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

--------------------------------------------------------------------------------
SQL handle: SQL_7aa929320eda7c96
SQL text: select /*+ full(a) */ * from scott.zkm a where object_id=1000
--------------------------------------------------------------------------------

--------------------------------------------------------------------------------
Plan name: SQL_PLAN_7pa99687dnz4q54f02d62         Plan id: 1425026402
Enabled: YES     Fixed: NO      Accepted: YES     Origin: MANUAL-LOAD
--------------------------------------------------------------------------------

Plan hash value: 3532417104

---------------------------------------------------------------------------------------------
| Id  | Operation                   | Name          | Rows  | Bytes | Cost (%CPU)| Time     |
---------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |               |  1097 |   221K|     7   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| ZKM           |  1097 |   221K|     7   (0)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN          | IDX_OBJECT_ID |   439 |       |     1   (0)| 00:00:01 |
---------------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   2 - access("OBJECT_ID"=1000)

25 rows selected.

Elapsed: 00:00:00.04

將原來的基線刪除，

17:49:14 SYS@zkm(1)> DECLARE
17:49:14   2   k1 pls_integer;
17:49:14   3  begin
17:49:14   4  k1 := DBMS_SPM.drop_sql_plan_baseline (  sql_handle=>'SQL_7aa929320eda7c96',plan_name=>'SQL_PLAN_7pa99687dnz4q45324a8c');
17:49:14   5  end;
17:49:14   6  /

PL/SQL procedure successfully completed.

Elapsed: 00:00:00.01
17:49:14 SYS@zkm(1)> col SQL_HANDLE for a25
17:49:19 SYS@zkm(1)> col plan_name for a35
17:49:19 SYS@zkm(1)> select signature,sql_handle,plan_name,origin,enabled,accepted,autopurge from dba_sql_plan_baselines;

 SIGNATURE SQL_HANDLE                PLAN_NAME                           ORIGIN                                     ENABLED   ACCEPTED  AUTOPURGE
---------- ------------------------- ----------------------------------- ------------------------------------------ --------- --------- ---------
8.8386E+18 SQL_7aa929320eda7c96      SQL_PLAN_7pa99687dnz4q54f02d62      MANUAL-LOAD                                YES       YES       YES

Elapsed: 00:00:00.00

再次執行sql，已經走索引了。

17:50:01 SYS@zkm(1)> select /*+ full(a) */ * from scott.zkm a where object_id=1000;

Elapsed: 00:00:00.00

Execution Plan
----------------------------------------------------------
Plan hash value: 3532417104

---------------------------------------------------------------------------------------------
| Id  | Operation                   | Name          | Rows  | Bytes | Cost (%CPU)| Time     |
---------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |               |  1097 |   221K|     7   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| ZKM           |  1097 |   221K|     7   (0)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN          | IDX_OBJECT_ID |   439 |       |     1   (0)| 00:00:01 |
---------------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   2 - access("OBJECT_ID"=1000)

Note
-----
   - SQL plan baseline "SQL_PLAN_7pa99687dnz4q54f02d62" used for this statement


Statistics
----------------------------------------------------------
         12  recursive calls
         10  db block gets
         78  consistent gets
          0  physical reads
          0  redo size
       1631  bytes sent via SQL*Net to client
        523  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
          0  sorts (memory)
          0  sorts (disk)
          1  rows processed

回退

把基線刪除即可。

參考

Oracle SQL執行計劃基線總結（SQL Plan Baseline）

Oracle如何固定執行計劃

相關推薦

Oracle使用SPM優化指定SQL

Oracle使用SPM優化指定SQL 環境構造 16:44:23 SYS@zkm(1)> drop table scott.zkm purge; Table dropped.

資料庫索引的優化及SQL處理過程

想要設計出好的索引，首先必須瞭解SQL語句在資料庫伺服器中的處理過程，本文介紹資料庫索引設計與優化中幾個對索引優化非常重要的概念。

11.mysql SQL優化之SQL問題定位

3.1 檢視SQL執行頻率 MySQL 客戶端連線成功後，通過 show [session|global] status 命令可以提供伺服器狀態資訊。show [session|global] status 可以根據需要加上引數“session”或者“global”

程式碼優化與sql優化---未完待續

　　萬丈高樓平地起，還是得打一個好地基呀 減少對變數對重複計算 //一般這麼寫

Mysql索引，有哪幾種索引，什麼時候該（不該）建索引；SQL怎麼進行優化以及SQL關鍵字的執行順序

索引（Index）是幫助MySQL高效獲取資料的資料結構。提取句子主幹，就可以得到索引的本質：索引是資料結構。

SQL基礎操作（十四）-JDBC連線優化、SQL語句測試

10.6 Statement物件 　　JDBC中的statement物件用於向資料庫傳送SQL語句，想完成資料的增刪改查，只需要通過這個物件向資料庫傳送增刪改查語句即可。

ORACLE效能優化之SQL語句優化

操作環境:AIX +11g+PLSQL 包含以下內容： SQL語句執行過程 優化器及執行計劃 合理應用Hints

阿里 P8 談：收穫不止 SOL 優化抓住 SQL 的本質，帶你領略 SQL 的世界！

開頭我先說： 有人就有江湖，有江湖就有 IT 系統，有 IT 系統就有資料庫，有資料庫就有 SQL，SQL 應用可一字概括:“\"廣\"。加之其簡單易學,SQL 實現也可一字概括:“樂”。

那些年我們一起優化的SQL https://mp.weixin.qq.com/s/sPO-6ULwIfUexLY3V4acBg

那些年我們一起優化的SQL https://mp.weixin.qq.com/s/sPO-6ULwIfUexLY3V4acBg 那些年我們一起優化的SQL

SQL優化一(SQL使用技巧)

1、行列轉換：   decode(條件,值1,返回值1,值2,返回值2,...值n,返回值n,預設值);   select decode(sign(變數1-變數2),-1,變數1,變數2) from dual; --取較小值

位元組跳動在Spark SQL上的核心優化實踐 | 位元組跳動技術沙龍

10月26日，位元組跳動技術沙龍 | 大資料架構專場 在上海位元組跳動總部圓滿結束。我們邀請到位元組跳動資料倉庫架構負責人-郭俊，Kyligence 大資料研發工程師-陶加濤，位元組跳動儲存工程師-徐明敏，阿里雲高階技術

T-sql指令碼級別優化

一概述 sql語句的執行時間一般會主要花在3步：語句編譯，語句執行，結果集返回.

淺談關於SQL優化的思路

零、為什麼要優化 系統的吞吐量瓶頸往往出現在資料庫的訪問速度上 隨著應用程式的執行，資料庫的中的資料會越來越多，處理時間會相應變慢

批量插入資料的優化，試試直接Append出SQL？

背景 最近在專案中碰到了需要批量插入資料的場景，當資料量在20w條的時候，就要花近36s（有索引）

MySQL千萬級大資料SQL查詢優化知識點總結

1.對查詢進行優化，應儘量避免全表掃描，首先應考慮在 where 及 order by 涉及的列上建立索引。

PL/SQL登入Oracle資料庫報錯ORA-12154:TNS:無法解析指定的連線識別符號已解決（本地未安裝Oracle需要連線伺服器上的）

ORA-12154:TNS:無法解析指定的連線識別符號 ORA-12154:TNS:無監聽程式 錯誤分析一、PL/SQL 客戶端登入到資料庫，如果配置錯誤會有以上錯誤，如下圖。

Mysql巧用join優化sql的方法詳解

0. 準備相關表來進行接下來的測試 相關建表語句請看：https://github.com/YangBaohust/my_sql

你真的知道怎麼優化SQL嗎

簡介 雖然使用Explain不能夠馬上調優我們的SQL，它也不能給予我們一些調整建議，但是它能夠讓我們瞭解MySQL 優化器是如何執行SQL 語句的

SQL語句優化的一些必會指南

前言 怎麼加快查詢速度，優化查詢效率，主要原則就是應儘量避免全表掃描，應該考慮在where及order by 涉及的列上建立索引。

資料庫sql語句優化

為什麼要優化： 隨著實際專案的啟動，資料庫經過一段時間的執行，最初的資料庫設定，會與實際資料庫執行效能會有一些差異，這時我們 就需要做一個優化調整。