oracle insert、append、parallel、随后查询的redo与磁盘读写
SQL> set autotrace traceonly statistics;
SQL> insert into big_table_dir_test1 select * from big_table_dir_test; 2853792 rows created. Statistics
----------------------------------------------------------
148 recursive calls
358348 db block gets
111261 consistent gets
2 physical reads
333542568 redo size
832 bytes sent via SQL*Net to client
817 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
2 sorts (memory)
0 sorts (disk)
2853792 rows processed SQL> commit; Commit complete. SQL> select f.owner,f.object_name,f.data_object_id,count(1),max(f.data_object_id),min(f.object_id) from big_table_dir_test1 f
group by f.owner,f.object_name,f.data_object_id; 2 87653 rows selected. Statistics
----------------------------------------------------------
7 recursive calls
1 db block gets
41034 consistent gets
0 physical reads -- 传统路径insert只写buffer cache, redo保证重做
176 redo size
4428645 bytes sent via SQL*Net to client
64793 bytes received via SQL*Net from client
5845 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
87653 rows processed SQL> truncate table big_table_dir_test1; Table truncated.
SQL> insert /*+ append nologging */ into big_table_dir_test1 select * from big_table_dir_test; 2853792 rows created. Statistics
----------------------------------------------------------
228 recursive calls
44268 db block gets
42998 consistent gets
2 physical reads
376672 redo size
827 bytes sent via SQL*Net to client
841 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
2853792 rows processed SQL> SQL> commit; Commit complete. SQL> select f.owner,f.object_name,f.data_object_id,count(1),max(f.data_object_id),min(f.object_id) from big_table_dir_test1 f
group by f.owner,f.object_name,f.data_object_id;
2 87653 rows selected. Statistics
----------------------------------------------------------
5 recursive calls
1 db block gets
40831 consistent gets
40752 physical reads --直接路径插入后,不经过buffer cache
168 redo size
4413020 bytes sent via SQL*Net to client
64793 bytes received via SQL*Net from client
5845 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
87653 rows processed SQL> SQL> / 87653 rows selected. Statistics
----------------------------------------------------------
0 recursive calls
0 db block gets
40766 consistent gets
0 physical reads
0 redo size
4310178 bytes sent via SQL*Net to client
64793 bytes received via SQL*Net from client
5845 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
87653 rows processed
SQL> truncate table big_table_dir_test1; Table truncated. Elapsed: 00:00:00.62
SQL>
SQL> alter session enable parallel dml; Session altered. Elapsed: 00:00:00.00
SQL> SQL> insert /*+ parallel(c,4) */ into big_table_dir_test1 c select * from big_table_dir_test; 2853792 rows created. Elapsed: 00:00:03.69 Statistics
----------------------------------------------------------
13 recursive calls
2574 db block gets
43108 consistent gets
0 physical reads
119108 redo size -- insert中的parallel导致走了直接路径加载
830 bytes sent via SQL*Net to client
840 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
2853792 rows processed SQL> commit; Commit complete.
SQL> select f.owner,f.object_name,f.data_object_id,count(1),max(f.data_object_id),min(f.object_id) from big_table_dir_test1 f
group by f.owner,f.object_name,f.data_object_id; 2 87653 rows selected. Elapsed: 00:00:03.33 Statistics
----------------------------------------------------------
5 recursive calls
1 db block gets
40896 consistent gets
40752 physical reads -- 没有写buffer cache
168 redo size
4470876 bytes sent via SQL*Net to client
64793 bytes received via SQL*Net from client
5845 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
87653 rows processed
SQL> truncate table big_table_dir_test1; insert into big_table_dir_test1 select /*+ parallel(b 4) */ * from big_table_dir_test b;
Table truncated. Elapsed: 00:00:00.05
SQL> SQL> 2853792 rows created. Elapsed: 00:00:04.66 Statistics
----------------------------------------------------------
139 recursive calls
358365 db block gets
110606 consistent gets
2 physical reads
333527468 redo size
846 bytes sent via SQL*Net to client
840 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
1 sorts (memory)
0 sorts (disk)
2853792 rows processed SQL> select /*+ parallel(4) */f.owner,f.object_name,f.data_object_id,count(1),max(f.data_object_id),min(f.object_id) from big_table_dir_test1 f
group by f.owner,f.object_name,f.data_object_id; 2 87653 rows selected. Elapsed: 00:00:02.07 Statistics
----------------------------------------------------------
38 recursive calls
1 db block gets
41750 consistent gets
0 physical reads -- parallel走了buffer cache
176 redo size
4557551 bytes sent via SQL*Net to client
64793 bytes received via SQL*Net from client
5845 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
87653 rows processed
目前暂时无法做到直接路径加载同时满足不生成redo,同时又写一份到buffer cache,这只能依赖于操作系统缓存,但是过多的并发append会导Disk file operations I/O致等待事件。
This event is used to wait for disk file operations (for example, open, close, seek, and resize). It is also used for miscellaneous I/O operations such as block dumps and password file accesses.
Wait Time: The wait time is the actual time it takes to do the I/O
Parameter |
Description |
FileOperation |
Type of file operation |
fileno |
File identification number |
filetype |
Type of file (for example, log file, data file, and so on) |
我们知道操作系统在操作文件的时候,需要打开文件、关闭文件、定位文件位置等,当这些操作在进行的时候,Oracle就处于等待状态。
操作系统的这些文件操作可以划分如下:
1.file creation
2 file open
3 file resize
4 file deletion
5 file close
6 wait for all aio requests to finish
7 write verification
8 wait for miscellaneous io (ftp, block dump, passwd file)
9 read from snapshot files
oracle insert、append、parallel、随后查询的redo与磁盘读写的更多相关文章
- Oracle insert /*+ APPEND */原理解析
https://blog.csdn.net/xiaobluesky/article/details/50494101 关于insert /*+ append */我们需要注意以下三点: a.非归档模式 ...
- Oracle NoLogging Append 方式减少批量insert的redo_size
业务处理中,很多时候使用实表临时表处理中间结果,而实表的Insert操作缺省会记录redo log,针对此问题收集相关测试总结信息如下: [转] 常见dml.ddl语句使用nologging选项所生成 ...
- 关于insert /*+ append*/ 各种insert插入速度比较
来源于:http://www.cnblogs.com/rootq/archive/2009/02/11/1388043.html SQL> select count(*) from t;COUN ...
- insert /*+APPEND*/ 各种insert 插入速度比较
SQL> select count(*) from t;COUNT(*)----------5442048****************************SQL> alter ta ...
- 快速向表中插入大量数据Oracle中append与Nologging
来源于:http://blog.sina.com.cn/s/blog_61cd89f60102e7gi.html 当需要对一个非常大的表INSERT的时候,会消耗非常多的资源,因为update表的时候 ...
- Oracle中append与Nologging
快速向表中插入大量数据Oracle中append与Nologging 2017-05-05 / VIEWS: 304 来源于:http://blog.sina.com.cn/s/blog_61cd89 ...
- Oracle ROWNUM用法和分页查询总结(转)
[转载] Oracle的分页查询语句基本上可以按照本文给出的格式来进行套用. Oracle分页查询格式(一):http://yangtingkun.itpub.net/post/468/100278 ...
- Oracle DBA 必须掌握的 查询脚本:
Oracle DBA 必须掌握的 查询脚本: 0:启动与关闭 orcle 数据库的启动与关闭 1:连接数据库 2:数据库开启状态的实现步骤: 2-1:启动数据库 2- ...
- Oracle ROWNUM用法和分页查询总结
**************************************************************************************************** ...
随机推荐
- layui---表单验证
使用layui,使用它的表单验证也是比不可少的,下面就来总结下: <!-- 不用form 用div也可以 --> <form class="layui-form" ...
- axis2与cxf区别
1.CXF支持 WS-Addressing,WS-Policy, WS-RM, WS-Security和WS-I Basic Profile.Axis2不支持WS-Policy,但是承诺在下面的版本支 ...
- spring-boot(三) HowTo
Spring Boot How To 1. 简介 本章节将回答一些常见的"我该怎么做"类型的问题,这些问题在我们使用spring Boot时经常遇到.这绝不是一个详尽的列表,但它覆 ...
- word-break和word-wrap的使用和区别
问题起源: 中文是一个字就是一个单词,而英文字母要有一个空格才将他们分割为一个单词:文字换行没事,主要是英文 <!DOCTYPE html> <html> <head&g ...
- html 中a标签的问题(无反应,跳转,调用方法)
让超链接点击后不跳转,可以用href = "#",但是这个#就会锚点到页面最上边 点击链接后不跳转可以设置成 1.<a href="javascri ...
- 关于Vue懒加载问题
有关Vue懒加载其实并不是想象的那么难和复杂: 首先引入 import VueLazyLoad from 'vue-lazyload'; 其次是使用 Vue.use(VueLazyLoad,{ er ...
- Javascript实现MD5加密
/* * A JavaScript implementation of the RSA Data Security, Inc. MD5 Message * Digest Algorithm, as d ...
- vue构造器的内容
<!DOCTYPE html> <html> <head> <meta charset="utf-8"> <title> ...
- [No0000D1]WPF—TreeView无限极绑定集合形成树结构
1.如图所示:绑定树效果图 2.前台Xaml代码: <Window x:Class="WpfTest.MainWindow" xmlns="http://schem ...
- :after和:before 伪类
1 使用伪类画三角形 .div{ margin-top: 100px; margin-left: 100px; } .div:after{ content: ''; display:inline-bl ...