Insert 语句对 nologging 与 logging表 在不同场景下的优化
前言
前段时间报表数据库上有条insert sql语句,插入的大量数据,执行非常慢,需要对其进行分析优化。
分析步骤是在:ARCHIVE与NOARCHIVE模式下进行。
测试场景: 分别对表的常规插入
,表在append插入
,表在append + parallel插入
进行性能测试,得出结果。
环境准备
数据库版本 | 基础表 | nologging表 | logging表 |
---|---|---|---|
Oracle 11g | T1 | T2 | T3 |
#创建T1,T2,T3表
create table t1 as select * from dba_objects;
create table t2 as select * from dba_objects where 1=2;
create table t3 as select * from dba_objects where 1=2;
#往T1表插入数据
SQL> insert into t1 select * from t1;
72813 rows created.
SQL> /
145626 rows created.
SQL> /
291252 rows created.
SQL> select count(*) from t1;
COUNT(*)
----------
582504
#设置T2表为nologging属性
SQL> alter table t2 nologging;
Table altered.
数据库处于ARCHIVE时
常规插入
nologging 表T2
SQL> insert into t2 select * from t1;
commit;
582824 rows created.
Execution Plan
----------------------------------------------------------
Plan hash value: 3617692013
---------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------------
| 0 | INSERT STATEMENT | | 582K| 53M| 1455 (2)| 00:00:18 |
| 1 | LOAD TABLE CONVENTIONAL | T2 | | | | |
| 2 | TABLE ACCESS FULL | T1 | 582K| 53M| 1455 (2)| 00:00:18 |
---------------------------------------------------------------------------------
Statistics
----------------------------------------------------------
3345 recursive calls
46879 db block gets
27878 consistent gets
8269 physical reads
67752144 redo size
838 bytes sent via SQL*Net to client
784 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
3 sorts (memory)
0 sorts (disk)
582824 rows processed
SQL>
Commit complete.
耗费:67752144
redo size
logging 表T3
SQL> insert into t3 select * from t1;
commit;
582824 rows created.
Execution Plan
----------------------------------------------------------
Plan hash value: 3617692013
---------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------------
| 0 | INSERT STATEMENT | | 582K| 53M| 1455 (2)| 00:00:18 |
| 1 | LOAD TABLE CONVENTIONAL | T3 | | | | |
| 2 | TABLE ACCESS FULL | T1 | 582K| 53M| 1455 (2)| 00:00:18 |
---------------------------------------------------------------------------------
Statistics
----------------------------------------------------------
2860 recursive calls
46875 db block gets
27811 consistent gets
1 physical reads
67875992 redo size
829 bytes sent via SQL*Net to client
784 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
582824 rows processed
SQL>
Commit complete.
耗费:67875992
redo size
append 插入
nologging 表T2
SQL> insert /*+ append */ into t2 select * from t1;
commit;
582824 rows created.
Execution Plan
----------------------------------------------------------
ERROR:
ORA-12838: cannot read/modify an object after modifying it in parallel
SP2-0612: Error generating AUTOTRACE EXPLAIN report
Statistics
----------------------------------------------------------
2627 recursive calls
9324 db block gets
8832 consistent gets
0 physical reads
143436 redo size
824 bytes sent via SQL*Net to client
798 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
582824 rows processed
耗费:143436
redo size
logging 表T3
SQL> insert /*+ append */ into t3 select * from t1;
582824 rows created.
Execution Plan
----------------------------------------------------------
ERROR:
ORA-12838: cannot read/modify an object after modifying it in parallel
SP2-0612: Error generating AUTOTRACE EXPLAIN report
Statistics
----------------------------------------------------------
2627 recursive calls
9327 db block gets
8832 consistent gets
0 physical reads
68384900 redo size
822 bytes sent via SQL*Net to client
797 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
582824 rows processed
耗费:68384900
redo size
parallel + append 插入
nologging 表T2
SQL> alter session enable parallel dml;
insert /*+ append parallel(2) */ into t2 select * from t1;
commit;
Session altered.
SQL>
582824 rows created.
Execution Plan
----------------------------------------------------------
ERROR:
ORA-12838: cannot read/modify an object after modifying it in parallel
SP2-0612: Error generating AUTOTRACE EXPLAIN report
Statistics
----------------------------------------------------------
52 recursive calls
32 db block gets
19 consistent gets
0 physical reads
21916 redo size
824 bytes sent via SQL*Net to client
809 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
582824 rows processed
耗费:21916
redo size
logging 表T3
SQL> alter session enable parallel dml;
insert /*+ append parallel(2)*/ into t3 select * from t1;
commit;
Session altered.
SQL>
582824 rows created.
Execution Plan
----------------------------------------------------------
ERROR:
ORA-12838: cannot read/modify an object after modifying it in parallel
SP2-0612: Error generating AUTOTRACE EXPLAIN report
Statistics
----------------------------------------------------------
50 recursive calls
33 db block gets
20 consistent gets
0 physical reads
21308 redo size
824 bytes sent via SQL*Net to client
808 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
582824 rows processed
耗费:21308
redo size
数据库处于NOARCHIVE时
常规插入
nologging 表T2
Execution Plan
----------------------------------------------------------
Plan hash value: 3617692013
---------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------------
| 0 | INSERT STATEMENT | | 582K| 53M| 1455 (2)| 00:00:18 |
| 1 | LOAD TABLE CONVENTIONAL | T2 | | | | |
| 2 | TABLE ACCESS FULL | T1 | 582K| 53M| 1455 (2)| 00:00:18 |
---------------------------------------------------------------------------------
Statistics
----------------------------------------------------------
2538 recursive calls
46869 db block gets
27796 consistent gets
8266 physical reads
67754744 redo size
824 bytes sent via SQL*Net to client
784 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
582824 rows processed
耗费:67754744
redo size
logging 表T3
Execution Plan
----------------------------------------------------------
Plan hash value: 3617692013
---------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------------
| 0 | INSERT STATEMENT | | 582K| 53M| 1455 (2)| 00:00:18 |
| 1 | LOAD TABLE CONVENTIONAL | T3 | | | | |
| 2 | TABLE ACCESS FULL | T1 | 582K| 53M| 1455 (2)| 00:00:18 |
---------------------------------------------------------------------------------
Statistics
----------------------------------------------------------
2593 recursive calls
46873 db block gets
27800 consistent gets
1600 physical reads
67757328 redo size
824 bytes sent via SQL*Net to client
784 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
3 sorts (memory)
0 sorts (disk)
582824 rows processed
耗费:67757328
redo size
append 插入
nologging 表T2
Statistics
----------------------------------------------------------
2627 recursive calls
9324 db block gets
8832 consistent gets
2993 physical reads
143480 redo size
822 bytes sent via SQL*Net to client
798 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
582824 rows processed
耗费:143480
redo size
logging 表T3
Statistics
----------------------------------------------------------
2627 recursive calls
9327 db block gets
8832 consistent gets
0 physical reads
143420 redo size
821 bytes sent via SQL*Net to client
798 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
582824 rows processed
耗费:143420
redo size
parallel + append 插入
nologging 表T2
Statistics
----------------------------------------------------------
50 recursive calls
32 db block gets
21 consistent gets
0 physical reads
21896 redo size
823 bytes sent via SQL*Net to client
810 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
582824 rows processed
耗费:21896
redo size
logging 表T3
Statistics
----------------------------------------------------------
50 recursive calls
33 db block gets
20 consistent gets
0 physical reads
21896 redo size
821 bytes sent via SQL*Net to client
809 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
582824 rows processed
耗费:21896
redo size
综合比较
属性 | 表名 | 常规插入产生的redo size | apppend插入产生的redo size | apppend + parallel插入产生的redo size |
---|---|---|---|---|
数据库模式 | archive | |||
nologing | t2 | 67752144 | 143436 | 21916 |
loging | t3 | 67875992 | 68384900 | 21308 |
数据库模式 | noarchive | |||
nologing | t2 | 67754744 | 143480 | 21896 |
loging | t3 | 67757328 | 143420 | 21896 |
1)数据库处于ARCHIVE模式时,
对logging表执行append插入,是对性能没有优化的
。加并行parallel
才会有影响。
2)数据库处于NOARCHIVE模式时,对logging表执行append插入,可以有效的提升性能。当然加并行parallel效果会更好
Insert 语句对 nologging 与 logging表 在不同场景下的优化的更多相关文章
- oracle带条件的Insert语句
背景 在一条记录完结时,自动向表中加入一条新的记录,采用的是事务处理,修改现有记录,并新增一条记录,直接采用的insert语句会报错 //主键冲突 unique constraint (XXXXXX) ...
- SQL基础语法—insert语句
1 insert语句 insert语句用于插入数据到表中,其基本语法有以下三种: Syntax: INSERT [LOW_PRIORITY | DELAYED | HIGH_PRIORITY] [IG ...
- 关于加快INSERT语句执行速度和HINT /*+ append */及/*+ append nologging */的使用
(非归档模式下)创建表T01: SQL> create table t01 as select * from dba_objects where 1=2; Table created. (非归档 ...
- SQLServer将表数据导出为Insert语句
从网上找到的方法,不过很不错,记录下来,也算是分享下~~ 有一个表,city,有列:cityID,cityName;将此表中所有数据,变为insert语句 select 'insert into ta ...
- 取得表中数据的insert语句
Build Insert Statements for the Existing Data in Tables 下面这个脚本实现了取得一个非空表中的所有insert语句 This script bui ...
- 【SQL Sever】将SQL Sever中的一个数据表的数据导出为insert语句
例如:这SQL Sever中的一张数据表,想要将这张数据表中的数据 转化成一个一个的insert语句存储在txt的文档中,那么不论走到那里这个insert语句一执行,我们就能将这个数据表中的数据 ...
- 使用C#导出MSSQL表数据Insert语句,支持所有MSSQL列属性
在正文开始之前,我们先看一下MSSQL的两张系统表sys.objects . syscolumnsMSDN中 sys.objects表的定义:在数据库中创建的每个用户定义的架构作用域内的对象在该表中均 ...
- select into from和insert into select from两种表复制语句区别
select into from和insert into select from两种表复制语句都是将源表source_table的记录插入到目标表target_table,但两句又有区别. 第一句(s ...
- 表数据转换为insert语句
/* 对象:导出物理表数据为Insert语句 描述:可以传递条件精确导出sql 加条件的前提是只知道相应的字段名及类型 */ from sysobjects where name ='proc_ins ...
随机推荐
- bzoj 1230: [Usaco2008 Nov]lites 开关灯【线段树】
在线段树上记录长度.区间01翻转标记.当前1个数.传递tag的时候用长度-1个数即可 #include<iostream> #include<cstdio> using nam ...
- bzoj 2142: 礼物【中国剩余定理+组合数学】
参考:http://blog.csdn.net/wzq_qwq/article/details/46709471 首先推组合数,设sum为每个人礼物数的和,那么答案为 \[ ( C_{n}^{sum} ...
- codeforces912E(折半搜索+双指针+二分答案)
E. Prime Gift E. Prime Gift time limit per test 3.5 seconds memory limit per test 256 megabytes inpu ...
- web项目tomcat启动url自定义(去掉项目名)
通常,使用maven构建web项目,启动时默认的访问路径: http://ip:port/项目名 很多时候我们不喜欢这样 访问,我们希望下面的访问方式: http://ip:port 如果是本地的to ...
- [SRM613~] TaroCheckers
一定要注意Topcoder的提交机制 Links: 原题地址 Vjudge Solution 这道题思维比较巧妙. 一看就基本知道是一个Dp题. 首先转换一下,用列而不是行来设第一维的状态,因为每列只 ...
- [Usaco2012 Nov]Concurrently Balanced Strings
Description [Brian Dean, 2012] Farmer John's cows are all of a very peculiar breed known for its dis ...
- [poj3744] Scout YYF I【概率dp 数学期望】
传送门:http://poj.org/problem?id=3744 令f(i)表示到i,安全的概率.则f(i) = f(i - 1) * p + f(i - 2) * (1 - p),若i位置有地雷 ...
- C. Mahmoud and a Message dp + 暴力
http://codeforces.com/contest/766/problem/C 关键在于dp,如何计数. 设dp[i]表示前i个字母中,能分成多少份合法的情况.那么答案就是dp[n],其中dp ...
- SpringBoot2.1.3修改tomcat参数支持请求特殊符号
最近遇到一个问题,比如GET请求中,key,value中带有特殊符号,请求会报错,见如下URL: http://xxx.xxx.xxx:8081/aaa?key1=val1&a.[].id=1 ...
- leetcode764 Largest Plus Sign
思路: 首先使用dp计算出在每个位置(i, j)上下左右最多有多少个连续的1,得到up[i][j], down[i][j], left[i][j], right[i][j].然后计算这四个值中的最小值 ...