首页 > 数据库 > 文库 > 正文

按 user 分组统计视图 全方位认识 sys 系统库

2024-09-07 22:12:41
字体:
来源:转载
供稿:网友

        在 《按 host 分组统计视图|全方位认识 sys 系统库》 中,我们介绍了sys 系统库中按 host 分组统计的视图,类似地,本期的内容将为大家介绍按照 user 进行分类统计的视图。下面请跟随我们一起开始 sys 系统库的系统学习之旅吧。
 
01
      user_summary,x$user_summary
      sys.x$memory_by_user_by_current_bytes
 
       下面我们看看使用该视图查询返回的结果。
 
# 不带x$前缀的视图
admin@localhost : sys 12:54:32> select * from user_summary limit 1/G
*************************** 1. row ***************************
              user: admin
        statements: 90530
statement_latency: 2.09 h
statement_avg_latency: 83.12 ms
      table_scans: 498
          file_ios: 60662
  file_io_latency: 31.05 s
current_connections: 4
total_connections: 1174
      unique_hosts: 2
    current_memory: 85.34 MiB
total_memory_allocated: 7.21 GiB
1 row in set (0.04 sec)
# 带x$前缀的视图
admin@localhost : sys 12:55:48> select * from x$user_summary limit 1/G
*************************** 1. row ***************************
              user: admin
        statements: 90752
statement_latency: 7524792139504000
statement_avg_latency: 82915992369.3583
      table_scans: 500
          file_ios: 60662
  file_io_latency: 31053125849250
current_connections: 4
total_connections: 1174
      unique_hosts: 2
    current_memory: 89381384
total_memory_allocated: 7755173436
1 row in set (0.02 sec)
视图字段含义如下:
 
user:客户端访问用户名。如果在performance_schema表中user列为NULL,则假定为后台线程,该字段为'background',如果为前台线程,则该字段对应具体的用户名
statements:对应用户执行的语句总数量
statement_latency:对应用户执行的语句总延迟时间(执行时间)
statement_avg_latency:对应用户执行的语句中,平均每个语句的延迟时间(执行时间)(SUM(stmt.total_latency/SUM(stmt.total))
table_scans:对应用户执行的语句发生表扫描总次数
file_ios:对应用户执行的语句产生的文件I/O事件总次数
file_io_latency:对应用户执行的语句产生的文件I/O事件的总延迟时间(执行时间)
current_connections:对应用户的当前连接数
total_connections:对应用户的历史总连接数
unique_hosts:对应用户来自不同主机(针对主机名去重)连接的数量
current_memory:对应用户的连接当前已使用的内存分配量
total_memory_allocated:对应用户的连接的历史内存分配量
PS:该视图只统计文件IO等待事件信息("wait/io/file/%")
 
02
user_summary_by_file_io,x$user_summary_by_file_io
按照用户分组的文件I/O延迟时间、IOS统计信息,默认按照总文件I/O时间延迟时间(执行时间)降序排序。数据来源:performance_schema.events_waits_summary_by_user_by_event_name
 
下面我们看看使用该视图查询返回的结果。
 
# 不带x$前缀的视图
admin@localhost : sys 12:56:18> select * from user_summary_by_file_io limit 3;
+------------+-------+------------+
| user      | ios  | io_latency |
+------------+-------+------------+
| admin      | 30331 | 15.53 s    |
| background | 10119 | 2.49 s    |
| qfsys      |  281 | 4.69 ms    |
+------------+-------+------------+
3 rows in set (0.01 sec)
# 带x$前缀的视图
admin@localhost : sys 12:56:21> select * from x$user_summary_by_file_io limit 3;
+------------+-------+----------------+
| user      | ios  | io_latency    |
+------------+-------+----------------+
| admin      | 30331 | 15526562924625 |
| background | 10122 |  2489231563125 |
| qfsys      |  281 |    4689150375 |
+------------+-------+----------------+
3 rows in set (0.00 sec)
视图字段含义如下:
 
user:客户端用户名。如果在performance_schema表中user列为NULL,则假定为后台线程,该字段为'background',如果为前台线程,则该字段对应具体的用户名
ios:对应用户的文件I/O事件总次数
io_latency:对应用户的文件I/O事件的总延迟时间(执行时间)
PS:该视图只统计文件IO等待事件信息("wait/io/file/%")
 
03
user_summary_by_file_io_type,x$user_summary_by_file_io_type
按照用户和事件类型(事件名称)分组的文件I/O延迟和IOS统计信息,默认情况下按照用户名和总文件I/O时间延迟时间(执行时间)降序排序。数据来源:performance_schema.events_waits_summary_by_user_by_event_name
 
下面我们看看使用该视图查询返回的结果。
 
# 不带x$前缀的视图
admin@localhost : sys 12:56:24> select * from user_summary_by_file_io_type limit 3;
+-------+-------------------------------------+-------+---------+-------------+
| user  | event_name                          | total | latency | max_latency |
+-------+-------------------------------------+-------+---------+-------------+
| admin | wait/io/file/sql/io_cache          | 27955 | 10.53 s | 67.61 ms    |
| admin | wait/io/file/innodb/innodb_log_file |  912 | 2.14 s  | 28.22 ms    |
| admin | wait/io/file/sql/binlog            |  879 | 2.05 s  | 31.75 ms    |
+-------+-------------------------------------+-------+---------+-------------+
3 rows in set (0.00 sec)
# 带x$前缀的视图
admin@localhost : sys 12:56:48> select * from x$user_summary_by_file_io_type limit 3;
+-------+-------------------------------------+-------+----------------+-------------+
| user  | event_name                          | total | latency        | max_latency |
+-------+-------------------------------------+-------+----------------+-------------+
| admin | wait/io/file/sql/io_cache          | 27955 | 10534662677625 | 67608294000 |
| admin | wait/io/file/innodb/innodb_log_file |  912 |  2143870695375 | 28216455000 |
| admin | wait/io/file/sql/binlog            |  879 |  2054976453000 | 31745275125 |
+-------+-------------------------------------+-------+----------------+-------------+
3 rows in set (0.01 sec)
视图字段含义如下:
 
user:客户端用户名。如果在performance_schema表中user列为NULL,则假定为后台线程,该字段为'background',如果为前台线程,则该字段对应具体的用户名
EVENT_NAME:文件I/O事件名称
total:对应用户发生的文件I/O事件总次数
latency:对应用户的文件I/O事件的总延迟时间(执行时间)
max_latency:对应用户的单次文件I/O事件的最大延迟时间(执行时间)
PS:该视图只统计文件IO等待事件信息("wait/io/file/%")
 
04
user_summary_by_stages,x$user_summary_by_stages
按用户分组的阶段事件统计信息,默认情况下按照用户名和阶段事件总延迟时间(执行时间)降序排序。数据来源:performance_schema.events_stages_summary_by_user_by_event_name
 
下面我们看看使用该视图查询返回的结果。
 
# 不带x$前缀的视图
admin@localhost : sys 12:56:51> select * from user_summary_by_stages limit 3;
+------------+-------------------------------+-------+---------------+-------------+
| user      | event_name                    | total | total_latency | avg_latency |
+------------+-------------------------------+-------+---------------+-------------+
| background | stage/innodb/buffer pool load |    1 | 12.56 s      | 12.56 s    |
+------------+-------------------------------+-------+---------------+-------------+
1 row in set (0.01 sec)
# 带x$前缀的视图
admin@localhost : sys 12:57:10> select * from x$user_summary_by_stages limit 3;
+------------+-------------------------------+-------+----------------+----------------+
| user      | event_name                    | total | total_latency  | avg_latency    |
+------------+-------------------------------+-------+----------------+----------------+
| background | stage/innodb/buffer pool load |    1 | 12561724877000 | 12561724877000 |
+------------+-------------------------------+-------+----------------+----------------+
1 row in set (0.00 sec)
视图字段含义如下:
 
user:客户端用户名。如果在performance_schema表中user列为NULL,则假定为后台线程,该字段为'background',如果为前台线程,则该字段对应具体的用户名
EVENT_NAME:阶段事件名称
total:对应用户的阶段事件的总次数
total_latency:对应用户的阶段事件的总延迟时间(执行时间)
avg_latency:对应用户的阶段事件的平均延迟时间(执行时间)
05
user_summary_by_statement_latency,x$user_summary_by_statement_latency
按照用户分组的语句统计信息,默认情况下按照语句总延迟时间(执行时间)降序排序。数据来源:performance_schema.events_statements_summary_by_user_by_event_name
 
下面我们看看使用该视图查询返回的结果。
 
# 不带x$前缀的视图
admin@localhost : sys 12:57:13> select * from user_summary_by_statement_latency limit 3;
+------------+-------+---------------+-------------+--------------+-----------+---------------+---------------+------------+
| user      | total | total_latency | max_latency | lock_latency | rows_sent | rows_examined | rows_affected | full_scans |
+------------+-------+---------------+-------------+--------------+-----------+---------------+---------------+------------+
| admin      | 45487 | 1.05 h        | 45.66 m    | 19.02 s      |      6065 |      17578842 |          1544 |        258 |
| qfsys      |    9 | 929.43 ms    | 928.68 ms  | 0 ps        |        5 |            0 |            0 |          0 |
| background |    0 | 0 ps          | 0 ps        | 0 ps        |        0 |            0 |            0 |          0 |
+------------+-------+---------------+-------------+--------------+-----------+---------------+---------------+------------+
3 rows in set (0.00 sec)
# 带x$前缀的视图
admin@localhost : sys 12:57:34> select * from x$user_summary_by_statement_latency limit 3;
+------------+-------+------------------+------------------+----------------+-----------+---------------+---------------+------------+
| user      | total | total_latency    | max_latency      | lock_latency  | rows_sent | rows_examined | rows_affected | full_scans |
+------------+-------+------------------+------------------+----------------+-----------+---------------+---------------+------------+
| admin      | 45562 | 3762457232413000 | 2739502018445000 | 19019928000000 |      6068 |      17579421 |          1544 |        259 |
| qfsys      |    9 |    929429421000 |    928682487000 |              0 |        5 |            0 |            0 |          0 |
| background |    0 |                0 |                0 |              0 |        0 |            0 |            0 |          0 |
+------------+-------+------------------+------------------+----------------+-----------+---------------+---------------+------------+
3 rows in set (0.00 sec)
视图字段含义如下:
 
user:客户端用户名。如果在performance_schema表中user列为NULL,则假定为后台线程,该字段为'background',如果为前台线程,则该字段对应具体的用户名
total:对应用户执行的语句总数量
total_latency:对应用户执行的语句总延迟时间(执行时间)
max_latency:对应用户执行的语句单次最大延迟时间(执行时间)
lock_latency:对应用户执行的语句锁等待的总时间
rows_sent:对应用户执行的语句返回给客户端的总数据行数
rows_examined:对应用户执行的语句从存储引擎读取的总数据行数
rows_affected:对应用户执行的语句影响的总数据行数
full_scans:对应用户执行的语句的全表扫描总次数
06
user_summary_by_statement_type,x$user_summary_by_statement_type
按用户和语句事件类型(事件类型名称为语句事件的event_name截取最后一部分字符串,也是语句command类型字符串类似)分组的语句统计信息,默认情况下按照用户名和对应语句的总延迟时间(执行时间)降序排序。数据来源:performance_schema.events_statements_summary_by_user_by_event_name
 
下面我们看看使用该视图查询返回的结果。
 
# 不带x$前缀的视图
admin@localhost : sys 12:57:38> select * from user_summary_by_statement_type limit 3;
+-------+-------------+-------+---------------+-------------+--------------+-----------+---------------+---------------+------------+
| user  | statement  | total | total_latency | max_latency | lock_latency | rows_sent | rows_examined | rows_affected | full_scans |
+-------+-------------+-------+---------------+-------------+--------------+-----------+---------------+---------------+------------+
| admin | alter_table |    2 | 56.56 m      | 43.62 m    | 0 ps        |        0 |            0 |            0 |          0 |
| admin | select      |  3662 | 5.53 m        | 2.02 m      | 4.73 s      |      6000 |      17532984 |            0 |        148 |
| admin | insert      |  1159 | 36.04 s      | 337.22 ms  | 14.23 s      |        0 |            0 |          1159 |          0 |
+-------+-------------+-------+---------------+-------------+--------------+-----------+---------------+---------------+------------+
3 rows in set (0.00 sec)
# 带x$前缀的视图
admin@localhost : sys 12:57:50> select * from x$user_summary_by_statement_type limit 3;
+-------+-------------+-------+------------------+------------------+----------------+-----------+---------------+---------------+------------+
| user  | statement  | total | total_latency    | max_latency      | lock_latency  | rows_sent | rows_examined | rows_affected | full_scans |
+-------+-------------+-------+------------------+------------------+----------------+-----------+---------------+---------------+------------+
| admin | alter_table |    2 | 3393877088372000 | 2617456143674000 |              0 |        0 |            0 |            0 |          0 |
| admin | select      |  3663 |  331756087959000 |  121243627173000 |  4733109000000 |      6003 |      17533557 |            0 |        149 |
| admin | insert      |  1159 |  36041502943000 |    337218573000 | 14229439000000 |        0 |            0 |          1159 |          0 |
+-------+-------------+-------+------------------+------------------+----------------+-----------+---------------+---------------+------------+
3 rows in set (0.00 sec)
视图字段含义如下:
 
user:客户端用户名。如果在performance_schema表中user列为NULL,则假定为后台线程,该字段为'background',如果为前台线程,则该字段对应具体的用户名
statement:语句事件名称的最后一部分字符串,与语句的command类型字符串类似
其他字段含义与 user_summary_by_statement_latency,x$user_summary_by_statement_latency 视图的字段含义相同。

(编辑:武林网)

发表评论 共有条评论
用户名: 密码:
验证码: 匿名发表