LOAD DATA INFILE – performance case study
转:
http://venublog.com/2007/11/07/load-data-infile-performance/
I often noticed that people complain about the LOAD DATA performance when loading the table with large number of rows of data. Even today I saw a case where the LOAD DATA on a simple 3 column table with about 5 million rows taking ~15 minutes of time. This is because the server did not had any tuning in regards to bulk insertion.
Consider the following simple MyISAM table on Redhat Linux 32-bit.
Shell
|
1
2
3
4
5
6
7
8
|
CREATE TABLE load1 (
`col1` varchar(100) NOT NULL default '',
`col2` int(11) default NULL,
`col3` char(1) default NULL,
PRIMARY KEY (`col1`)
) TYPE=MyISAM;
|
The table has a string key column. Here is the data file(download here) that I used it for testing:
Shell
|
1
2
3
4
5
6
7
|
[vanugant@escapereply:t55 tmp]$ wc loaddata.csv
5164946 5164946 227257389 loaddata.csv
[vanugant@escapereply:t55 tmp]$ ls -alh loaddata.csv
-rw-r--r-- 1 vanugant users 217M Nov 6 14:42 loaddata.csv
[vanugant@escapereply:t55 tmp]$
|
Here is the default mysql system variables related to LOAD DATA:
Shell
|
1
2
3
4
5
6
7
8
9
10
|
mysql> show variables;
+-------------------------+---------+
| Variable_name | Value |
+-------------------------+---------+
| bulk_insert_buffer_size | 8388608 |
| myisam_sort_buffer_size | 16777216 |
| key_buffer_size | 33554432 |
+-------------------------+----------+
|
and here is the actual LOAD DATA query to load all ~5m rows (~256M of data) to the table and its timing.
Shell
|
1
2
3
4
5
|
mysql> LOAD DATA INFILE '/home/vanugant/tmp/loaddata.csv' IGNORE INTO TABLE load1 FIELDS TERMINATED BY ',';
Query OK, 4675823 rows affected (14 min 56.84 sec)
Records: 5164946 Deleted: 0 Skipped: 489123 Warnings: 0
|
Now, lets experiment by disabling the keys in the table before running the LOAD DATA:
Shell
|
1
2
3
4
5
6
7
8
9
10
11
|
mysql> SET SESSION BULK_INSERT_BUFFER_SIZE=314572800;
Query OK, 0 rows affected (0.00 sec)
mysql> alter table load1 disable keys;
Query OK, 0 rows affected (0.00 sec)
mysql> LOAD DATA INFILE '/home/vanugant/tmp/loaddata.csv' IGNORE INTO TABLE load1 FIELDS TERMINATED BY ',';
Query OK, 4675823 rows affected (13 min 47.50 sec)
Records: 5164946 Deleted: 0 Skipped: 489123 Warnings: 0
|
No use, just 1% increase or same…., now lets set the real MyISAM values… and try again…
Shell
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
|
mysql> SET SESSION BULK_INSERT_BUFFER_SIZE=256217728;
Query OK, 0 rows affected (0.00 sec)
mysql> set session MYISAM_SORT_BUFFER_SIZE=256217728;
Query OK, 0 rows affected (0.00 sec)
mysql> set global KEY_BUFFER_SIZE=256217728;
Query OK, 0 rows affected (0.05 sec)
mysql> alter table load1 disable keys;
Query OK, 0 rows affected (0.00 sec)
mysql> LOAD DATA INFILE '/home/vanugant/tmp/loaddata.csv' IGNORE INTO TABLE load1 FIELDS TERMINATED BY ',';
Query OK, 4675823 rows affected (1 min 55.05 sec)
Records: 5164946 Deleted: 0 Skipped: 489123 Warnings: 0
mysql> alter table load1 enable keys;
Query OK, 0 rows affected (0.00 sec)
|
Wow…thats almost 90% increase in the performance. So, disabling the keys in MyISAM is not just the key, but tuning the buffer size does play role based on the input data.
For the same case with Innodb, here is the status by adjusting the Innodb_buffer_pool_size=1G andInnodb_log_file_size=256M along with innodb_flush_logs_at_trx_commit=1.
Shell
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
|
mysql> show variables like '%innodb%size';
+---------------------------------+------------+
| Variable_name | Value |
+---------------------------------+------------+
| innodb_additional_mem_pool_size | 26214400 |
| innodb_buffer_pool_size | 1073741824 |
| innodb_log_buffer_size | 8388608 |
| innodb_log_file_size | 268435456 |
+---------------------------------+------------+
mysql> LOAD DATA INFILE '/home/vanugant/tmp/loaddata.csv' IGNORE INTO TABLE load1 FIELDS TERMINATED BY ',';
Query OK, 4675823 rows affected (2 min 37.53 sec)
Records: 5164946 Deleted: 0 Skipped: 489123 Warnings: 0
|
With innodb_flush_logs_at_trx_commit=2, innodb_flush_method=O_DIRECT and innodb_doublewrite=0; it will be another 40% difference (use all these variables with caution, unless you know what you are doing)
Shell
|
1
2
3
4
5
|
mysql> LOAD DATA INFILE '/home/vanugant/tmp/loaddata.csv' IGNORE INTO TABLE load1 FIELDS TERMINATED BY ',';
Query OK, 4675823 rows affected (1 min 53.69 sec)
Records: 5164946 Deleted: 0 Skipped: 489123 Warnings: 0
|
LOAD DATA INFILE – performance case study的更多相关文章
- LOAD DATA INFILE Syntax--官方
LOAD DATA [LOW_PRIORITY | CONCURRENT] [LOCAL] INFILE 'file_name' [REPLACE | IGNORE] INTO TABLE tbl_n ...
- Data Visualization – Banking Case Study Example (Part 1-6)
python信用评分卡(附代码,博主录制) https://study.163.com/course/introduction.htm?courseId=1005214003&utm_camp ...
- Mysql load data infile 导入数据出现:Data truncated for column
[1]Mysql load data infile 导入数据出现:Data truncated for column .... 可能原因分析: (1)数据库表对应字段类型长度不够或修改为其他数据类型( ...
- Mysql load data infile 命令导入含中文csv源数据文件 【错误代码 1300】
[1]Load data infile 命令导入含中文csv源数据文件 报错:Invalid utf8 character string: '??֧' (1)问题现象 csv格式文件源数据: 导入SQ ...
- Mysql load data infile 命令格式
[1]Linux系统环境下 LOAD DATA INFILE /usr/LOCAL/lib/ubcsrvd/datacsv/201909_source.csv INTO TABLE np_cdr_20 ...
- Mysql 命令 load data infile 权限问题
[1]Mysql命令load data infile 执行权限问题 工作中,经常会遇到往线上环境mysql数据库批量导入源数据的场景. 针对这个场景问题,mysql有一个很高效的命令:load dat ...
- mysql load data infile的使用 和 SELECT into outfile备份数据库数据
LOAD DATA [LOW_PRIORITY | CONCURRENT] [LOCAL] INFILE 'file_name.txt' [REPLACE | IGNORE] INTO TABLE t ...
- SQL基本语句(3) LOAD DATA INFILE
使用LOAD语句批量录入数据 语法: LOAD DATA [LOW_PRIORITY | CONCURRENT] [LOCAL] INFILE 'file_name' [REPLACE | IGNOR ...
- mysql导入数据load data infile用法
mysql导入数据load data infile用法 基本语法: load data [low_priority] [local] infile 'file_name txt' [replace | ...
随机推荐
- Visual Studio 2010 快捷键
Visual Studio 2010 快捷键: CTRL + SHIFT + B 生成解决方案CTRL + F7 生成编译CTRL + O 打开文件CTRL + SHIFT + O 打开项目CTRL ...
- AcmeAir安装AI探针--SaaS版
一.安装SaaS版AI探针准备工作: 1.准备好可用的docker版AcmeAir应用 2.在SaaS官网tpm.oneapm.com上已注册可用的账号 3.登录tpm.oneapm.com,点击选择 ...
- 正确的理解this 和 super
this和super是Java的两个关键字. 先明确一个问题,有人错误的认为它们是对象里的“属性”,这只能怪老师没有讲清楚计算机的本质了.因为计算机的处理器只能用指令去处理数据,像C语言之类的容易理解 ...
- JVM加载class文件的原理
当Java编译器编译好.class文件之后,我们需要使用JVM来运行这个class文件.那么最开始的工作就是要把字节码从磁盘输入到内存中,这个过程我们叫做[加载 ].加载完成之后,我们就可以进行一系列 ...
- 统计MySQL数据表大小
SELECT CONCAT(TRUNCATE(SUM(data_length)/1024/1024,2),'MB') AS data_size,CONCAT(TRUNCATE(SUM(max_data ...
- c# 网络是否连接
c# 网络是否连接 方案一: using System; using System.Collections.Generic; using System.Linq; using System.Text ...
- perl install module as non-root user
install to local directory. 1. cpan 初始化,不用local::lib,mannual就行,其他auto2. 修改cpan 配置文件 cpan > o conf ...
- python 列表(list)去除重复的元素总结
方法一: 将list作为set的构造函数构造一个set,然后再将set转换会list就可以 >>> myList = [1, 2, 3, 3, 2, 2, 4, 5, 5] > ...
- Python 中 os.path模板
os.path.abspath(path) #返回绝对路径 os.path.basename(path) #返回文件名 os.path.commonprefix(list) #返回list(多个路径) ...
- WebDriverExtensionsByC#
测试工具//********************************************************************************************** ...