统计信息查询视图 | 全方位认识 sys 系统库

在《会话和锁信息查询视图|全方位认识 sys 系统库》中,为大家介绍了如何使用 sys 系统库总的视图来查询会话状态信息以及锁等待信息,本期内容将为大家介绍查询表和索引相关的统计信息快捷视图。

作者 罗小波·沃趣科技高级数据库技术专家
出品 沃趣科技

在上一篇《会话和锁信息查询视图|全方位认识 sys 系统库》中,我们介绍了如何使用 sys 系统库总的视图来查询会话状态信息以及锁等待信息,本期的内容先给大家介绍查询表和索引相关的统计信息快捷视图。下面请跟随我们一起开始 sys 系统库的系统学习之旅吧。

PS:由于本文中所提及的视图功能的特殊性(DBA日常工作中可能需要查询一些统计信息做一些数据分析使用),所以下文中会列出部分视图中的select语句文本,以便大家更直观地学习它们。

01.schema_auto_increment_columns

在所有数据库中(排除系统字典库 mysql,sys,INFORMATION_SCHEMA,performance_schema)查找带有自增列的基表及其相关的信息,默认按照自增值使用率和自增列类型最大值进行降序排序。数据来源:INFORMATION_SCHEMA的COLUMNS、TABLES

  • 此视图在MySQL 5.7.9中新增

视图查询语句文本

SELECT TABLE_SCHEMA,
  TABLE_NAME,
  COLUMN_NAME,
  DATA_TYPE,
  COLUMN_TYPE,
  (LOCATE('unsigned', COLUMN_TYPE) = 0) AS is_signed,
  (LOCATE('unsigned', COLUMN_TYPE) > 0) AS is_unsigned,
  (
      CASE DATA_TYPE
        WHEN 'tinyint' THEN 255
        WHEN 'smallint' THEN 65535
        WHEN 'mediumint' THEN 16777215
        WHEN 'int' THEN 4294967295
        WHEN 'bigint' THEN 18446744073709551615
      END >> IF(LOCATE('unsigned', COLUMN_TYPE) > 0, 0, 1)
  ) AS max_value,
  AUTO_INCREMENT,
  AUTO_INCREMENT / (
    CASE DATA_TYPE
      WHEN 'tinyint' THEN 255
      WHEN 'smallint' THEN 65535
      WHEN 'mediumint' THEN 16777215
      WHEN 'int' THEN 4294967295
      WHEN 'bigint' THEN 18446744073709551615
    END >> IF(LOCATE('unsigned', COLUMN_TYPE) > 0, 0, 1)
  ) AS auto_increment_ratio
FROM INFORMATION_SCHEMA.COLUMNS
INNER JOIN INFORMATION_SCHEMA.TABLES USING (TABLE_SCHEMA, TABLE_NAME)
WHERE TABLE_SCHEMA NOT IN ('mysql', 'sys', 'INFORMATION_SCHEMA', 'performance_schema')
AND TABLE_TYPE='BASE TABLE'
AND EXTRA='auto_increment'
ORDER BY auto_increment_ratio DESC, max_value;

下面我们看看使用该视图查询返回的结果

admin@localhost : sys 11:11:58> select * from schema_auto_increment_columns limit 5;
+--------------+------------+-------------+-----------+------------------+-----------+-------------+------------+----------------+----------------------+
| table_schema | table_name | column_name | data_type | column_type      | is_signed | is_unsigned | max_value  | auto_increment | auto_increment_ratio |
+--------------+------------+-------------+-----------+------------------+-----------+-------------+------------+----------------+----------------------+
| sbtest      | sbtest1    | id          | int      | int(10) unsigned |        0 |          1 | 4294967295 |      10713891 |              0.0025 |
| sbtest      | sbtest2    | id          | int      | int(10) unsigned |        0 |          1 | 4294967295 |      10710865 |              0.0025 |
| sbtest      | sbtest3    | id          | int      | int(10) unsigned |        0 |          1 | 4294967295 |      10714919 |              0.0025 |
| sbtest      | sbtest4    | id          | int      | int(10) unsigned |        0 |          1 | 4294967295 |      10714039 |              0.0025 |
| sbtest      | sbtest5    | id          | int      | int(10) unsigned |        0 |          1 | 4294967295 |      10713075 |              0.0025 |
+--------------+------------+-------------+-----------+------------------+-----------+-------------+------------+----------------+----------------------+
5 rows in set (1.50 sec)

视图字段含义如下:

  • TABLE_SCHEMA:包含自增值的表的schema名称

  • TABLE_NAME:包含AUTO_INCREMENT值的表名

  • column_name:AUTO_INCREMENT值的列名称

  • data_type:自增列的数据类型

  • COLUMN_TYPE:自增列的列属性类型,即在数据类型基础上加上一些其他信息。例如:对于bigint(20) unsigned,整个信息就被称为列属性类型,而数据类型只是指的bigint

  • is_signed:列类型是否是有符号的

  • is_unsigned:列类型是否是无符号的

  • MAX_VALUE:自增列的最大自增值

  • auto_increment:自增列的当前AUTO_INCREMENT属性值

  • auto_increment_ratio:自增列当前使用的自增值与自增列最大自增值的比例,表示当前自增列的使用率

02.schema_index_statistics,x$schema_index_statistics

索引统计信息,默认按照使用索引执行增删改查操作的总延迟时间(执行时间)降序排序,数据来源:performance_schema.table_io_waits_summary_by_index_usage

视图查询语句文本

# 不带x$前缀的视图
SELECT OBJECT_SCHEMA AS table_schema,
  OBJECT_NAME AS table_name,
  INDEX_NAME as index_name,
  COUNT_FETCH AS rows_selected,
  sys.format_time(SUM_TIMER_FETCH) AS select_latency,
  COUNT_INSERT AS rows_inserted,
  sys.format_time(SUM_TIMER_INSERT) AS insert_latency,
  COUNT_UPDATE AS rows_updated,
  sys.format_time(SUM_TIMER_UPDATE) AS update_latency,
  COUNT_DELETE AS rows_deleted,
  sys.format_time(SUM_TIMER_INSERT) AS delete_latency
FROM performance_schema.table_io_waits_summary_by_index_usage
WHERE index_name IS NOT NULL
ORDER BY sum_timer_wait DESC;

# 带x$前缀的视图查询语句与不带x$前缀的视图查询语句相比,只是少了单位格式化函数
......

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 11:19:43> select * from schema_index_statistics limit 5;
+--------------+------------+------------+---------------+----------------+---------------+----------------+--------------+----------------+--------------+----------------+
| table_schema | table_name | index_name | rows_selected | select_latency | rows_inserted | insert_latency | rows_updated | update_latency | rows_deleted | delete_latency |
+--------------+------------+------------+---------------+----------------+---------------+----------------+--------------+----------------+--------------+----------------+
| xiaoboluo    | test      | PRIMARY    |          1159 | 3.57 s        |            0 | 0 ps          |            0 | 0 ps          |            0 | 0 ps          |
| sys          | sys_config | PRIMARY    |            1 | 62.53 ms      |            0 | 0 ps          |            0 | 0 ps          |            0 | 0 ps          |
| sbtest      | sbtest1    | i_c        |            20 | 31.43 ms      |            0 | 0 ps          |            0 | 0 ps          |            0 | 0 ps          |
| xiaoboluo    | test      | i_test    |          400 | 3.77 ms        |            0 | 0 ps          |            0 | 0 ps          |            0 | 0 ps          |
| luoxiaobo    | public_num | PRIMARY    |            0 | 0 ps          |            0 | 0 ps          |            0 | 0 ps          |            0 | 0 ps          |
+--------------+------------+------------+---------------+----------------+---------------+----------------+--------------+----------------+--------------+----------------+
5 rows in set (0.45 sec)

# 带x$前缀的视图
admin@localhost : sys 11:20:21> select * from x$schema_index_statistics limit 5;
+--------------+------------+------------+---------------+----------------+---------------+----------------+--------------+----------------+--------------+----------------+
| table_schema | table_name | index_name | rows_selected | select_latency | rows_inserted | insert_latency | rows_updated | update_latency | rows_deleted | delete_latency |
+--------------+------------+------------+---------------+----------------+---------------+----------------+--------------+----------------+--------------+----------------+
| xiaoboluo    | test      | PRIMARY    |          1159 |  3573795058125 |            0 |              0 |            0 |              0 |            0 |              0 |
| sys          | sys_config | PRIMARY    |            1 |    62528964375 |            0 |              0 |            0 |              0 |            0 |              0 |
| sbtest      | sbtest1    | i_c        |            20 |    31429669125 |            0 |              0 |            0 |              0 |            0 |              0 |
| xiaoboluo    | test      | i_test    |          400 |    3765146625 |            0 |              0 |            0 |              0 |            0 |              0 |
| luoxiaobo    | public_num | PRIMARY    |            0 |              0 |            0 |              0 |            0 |              0 |            0 |              0 |
+--------------+------------+------------+---------------+----------------+---------------+----------------+--------------+----------------+--------------+----------------+
5 rows in set (0.00 sec)

视图字段含义如下:

  • TABLE_SCHEMA:包含索引的表所在的schema名称

  • TABLE_NAME:包含索引的表名

  • INDEX_NAME:索引的名称

  • rows_selected:使用索引读的总数据行数

  • select_latency:使用索引读的总延迟时间(执行时间)

  • rows_inserted:插入索引的总行数

  • insert_latency:插入索引行的总延迟时间(执行时间)

  • rows_updated:索引更新的总行数

  • update_latency:索引更新行的总延迟时间(执行时间)

  • rows_deleted:从索引中删除的总行数

  • delete_latency:从索引中删除行的总延迟时间(执行时间)

03.schema_object_overview

每个schema中包含的表、视图、索引等对象的统计信息,默认按照schema名称和对象类型进行排序,数据来源:information_schema的routines、tables、statistics、triggers、events

  • 注意:对于具有大量数据库对象的MySQL实例,此视图可能需要很长时间才能执行完成

视图查询语句文本

SELECT ROUTINE_SCHEMA AS db, ROUTINE_TYPE AS object_type, COUNT(*) AS count FROM information_schema.routines GROUP BY ROUTINE_SCHEMA, ROUTINE_TYPE
UNION
SELECT TABLE_SCHEMA, TABLE_TYPE, COUNT(*) FROM information_schema.tables GROUP BY TABLE_SCHEMA, TABLE_TYPE
UNION
SELECT TABLE_SCHEMA, CONCAT('INDEX (', INDEX_TYPE, ')'), COUNT(*) FROM information_schema.statistics GROUP BY TABLE_SCHEMA, INDEX_TYPE
UNION
SELECT TRIGGER_SCHEMA, 'TRIGGER', COUNT(*) FROM information_schema.triggers GROUP BY TRIGGER_SCHEMA
UNION
SELECT EVENT_SCHEMA, 'EVENT', COUNT(*) FROM information_schema.events GROUP BY EVENT_SCHEMA
ORDER BY DB, OBJECT_TYPE;

下面我们看看使用该视图查询返回的结果

admin@localhost : sys 11:20:27> select * from schema_object_overview limit 10;
+--------------------+---------------+-------+
| db                | object_type  | count |
+--------------------+---------------+-------+
| information_schema | SYSTEM VIEW  |    61 |
| luoxiaobo          | BASE TABLE    |    3 |
| luoxiaobo          | INDEX (BTREE) |    3 |
| mysql              | BASE TABLE    |    31 |
| mysql              | INDEX (BTREE) |    69 |
| performance_schema | BASE TABLE    |    87 |
| qfsys              | BASE TABLE    |    1 |
| qfsys              | INDEX (BTREE) |    1 |
| sbtest            | BASE TABLE    |    8 |
| sbtest            | INDEX (BTREE) |    17 |
+--------------------+---------------+-------+
10 rows in set (0.27 sec)

视图字段含义如下:

  • db:schema名称

  • OBJECT_TYPE:数据库对象类型,有效值为:BASE TABLE,INDEX(index_type),EVENT,FUNCTION,PROCEDURE,TRIGGER,VIEW

  • count:在每个schema下各个数据库对象的数量

04.schema_redundant_indexes

查找重复或冗余索引,数据来源:sys.x$schema_flattened_keys,该数据来源视图被称作schema_redundant_indexes视图的辅助视图

  • schema_redundant_indexes视图在MySQL 5.7.9中新增

下面我们看看使用该视图查询返回的结果

admin@localhost : sys 11:21:13> select * from schema_redundant_indexes limit 1\G;
*************************** 1. row ***************************
          table_schema: test
            table_name: test
  redundant_index_name: i_id
  redundant_index_columns: id
  redundant_index_non_unique: 1
  dominant_index_name: i_id_id2
dominant_index_columns: id,id2
dominant_index_non_unique: 1
        subpart_exists: 0
        sql_drop_index: ALTER TABLE `test`.`test` DROP INDEX `i_id`
1 row in set (0.01 sec)

视图字段含义如下:

  • TABLE_SCHEMA:包含冗余或重复索引的表对应的schema名称

  • TABLE_NAME:包含冗余或重复索引的表名

  • redundant_index_name:冗余或重复的索引名称

  • redundant_index_columns:冗余或重复索引中的列名

  • redundant_index_non_unique:冗余或重复索引中非唯一列的数量

  • dominant_index_name:与重复或冗余索引相比占据优势(最佳)的索引名称

  • dominant_index_columns:占据优势(最佳)的索引中的列名

  • dominant_index_non_unique:占据优势(最佳)的索引中非唯一列的数量

  • subpart_exists:重复或冗余索引是否是前缀索引

  • sql_drop_index:针对重复或冗余索引生成的drop index语句

05.schema_table_statistics,x$schema_table_statistics

查看表的统计信息,默认情况下按照增删改查操作的总表I/O延迟时间(执行时间,即也可以理解为是存在最多表I/O争用的表)降序排序,数据来源:performance_schema.table_io_waits_summary_by_table、sys.x$ps_schema_table_statistics_io

  • 这些视图使用了一个辅助视图x$ps_schema_table_statistics_io

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 11:52:25> select * from schema_table_statistics limit 1\G
*************************** 1. row ***************************
table_schema: xiaoboluo
  table_name: test
total_latency: 2.10 m
rows_fetched: 1561
fetch_latency: 2.08 m
rows_inserted: 1159
insert_latency: 865.33 ms
rows_updated: 0
update_latency: 0 ps
rows_deleted: 0
delete_latency: 0 ps
io_read_requests: 43
      io_read: 178.86 KiB
io_read_latency: 15.00 ms
io_write_requests: 10
    io_write: 160.00 KiB
io_write_latency: 76.24 us
io_misc_requests: 42
io_misc_latency: 9.38 ms
1 row in set (0.03 sec)

# 带x$前缀的视图
admin@localhost : sys 11:52:28> select * from x$schema_table_statistics limit 1\G;
*************************** 1. row ***************************
table_schema: xiaoboluo
  table_name: test
total_latency: 125711643303375
rows_fetched: 1561
fetch_latency: 124846318302750
rows_inserted: 1159
insert_latency: 865325000625
rows_updated: 0
update_latency: 0
rows_deleted: 0
delete_latency: 0
io_read_requests: 43
      io_read: 183148
io_read_latency: 15001512375
io_write_requests: 10
    io_write: 163840
io_write_latency: 76237125
io_misc_requests: 42
io_misc_latency: 9384933000
1 row in set (0.02 sec)

视图字段含义如下:

  • TABLE_SCHEMA:包含TABLE_NAME字段的表所在的schema名称

  • TABLE_NAME:表名

  • total_latency:表的I/O事件的总延迟时间(执行时间),针对表增删改查操作

  • rows_fetched:表读取操作的总数据行数,针对表查询操作

  • fetch_latency:表select操作的I/O事件的总延迟时间(执行时间),针对表查询操作

  • rows_inserted:表插入操作的总数据行数,针对表插入操作

  • insert_latency:表insert操作的I/O事件的延迟时间(执行时间),针对表插入操作

  • rows_updated:表更新操作的总数据行数,针对表更新操作

  • update_latency:表更新操作的I/O事件的总延迟时间(执行时间),针对表更新操作

  • rows_deleted:表删除操作的总数据行数,针对表删除操作

  • delete_latency:表删除操作的I/O事件的总延迟时间(执行时间),针对表删除操作

  • io_read_requests:表读取操作总请求次数,针对表.ibd和.frm文件的读I/O操作

  • io_read:表读操作相关的所有文件读取操作的总字节数,针对表.ibd和.frm文件的读I/O操作

  • io_read_latency:表读操作相关的所有文件读取操作的总延迟时间(执行时间),针对表.ibd和.frm文件的读I/O操作

  • io_write_requests:表写操作的总请求次数,针对表.ibd和.frm文件的写I/O操作

  • io_write:表写操作相关的所有文件写操作的总字节数,针对表.ibd和.frm文件的写I/O操作

  • io_write_latency:表写操作相关的所有文件写操作的总延迟时间(执行时间),针对表.ibd和.frm文件的写I/O操作

  • io_misc_requests:表其他各种混杂操作相关的所有文件的I/O请求总次数,针对表.ibd和.frm文件的其他混杂I/O操作

  • io_misc_latency:表其他各种混杂操作相关的所有文件的I/O请求的总延迟时间(执行时间),针对表.ibd和.frm文件的其他混杂I/O操作

06.schema_table_statistics_with_buffer,x$schema_table_statistics_with_buffer

查询表的统计信息,其中还包括InnoDB缓冲池统计信息,默认情况下按照增删改查操作的总表I/O延迟时间(执行时间,即也可以理解为是存在最多表I/O争用的表)降序排序,数据来源:performance_schema.table_io_waits_summary_by_table、sys.x$ps_schema_table_statistics_io、sys.x$innodb_buffer_stats_by_table

  • 这些视图使用了辅助视图sys.x$ps_schema_table_statistics_io

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 12:36:57> select * from schema_table_statistics_with_buffer limit 1\G;
*************************** 1. row ***************************
          table_schema: xiaoboluo
            table_name: test
          rows_fetched: 1561
        fetch_latency: 2.08 m
        rows_inserted: 1159
        insert_latency: 865.33 ms
          rows_updated: 0
        update_latency: 0 ps
          rows_deleted: 0
        delete_latency: 0 ps
      io_read_requests: 48
              io_read: 179.29 KiB
      io_read_latency: 15.02 ms
    io_write_requests: 10
              io_write: 160.00 KiB
      io_write_latency: 76.24 us
      io_misc_requests: 47
      io_misc_latency: 9.47 ms
    innodb_buffer_allocated: 112.00 KiB
    innodb_buffer_data: 48.75 KiB
    innodb_buffer_free: 63.25 KiB
  innodb_buffer_pages: 7
innodb_buffer_pages_hashed: 0
innodb_buffer_pages_old: 0
innodb_buffer_rows_cached: 1162
1 row in set (2.21 sec)

# 带x$前缀的视图
admin@localhost : sys 12:37:35> select * from x$schema_table_statistics_with_buffer limit 1\G;
*************************** 1. row ***************************
          table_schema: xiaoboluo
            table_name: test
          rows_fetched: 1561
        fetch_latency: 124846318302750
        rows_inserted: 1159
        insert_latency: 865325000625
          rows_updated: 0
        update_latency: 0
          rows_deleted: 0
        delete_latency: 0
      io_read_requests: 48
              io_read: 183595
      io_read_latency: 15019373250
    io_write_requests: 10
              io_write: 163840
      io_write_latency: 76237125
      io_misc_requests: 47
      io_misc_latency: 9465938250
innodb_buffer_allocated: 114688
    innodb_buffer_data: 49917
    innodb_buffer_free: 64771
  innodb_buffer_pages: 7
innodb_buffer_pages_hashed: 0
innodb_buffer_pages_old: 0
innodb_buffer_rows_cached: 1162
1 row in set (2.12 sec)

视图字段含义如下:

  • 表相关的统计信息字段的含义与视图schema_table_statistics的字段含义相同,这里省略,详见schema_table_statistics,x$schema_table_statistics 视图解释部分

  • innodb_buffer_allocated:当前已分配给表的buffer pool总字节数

  • innodb_buffer_data:当前已分配给表的数据部分使用的buffer pool字节总数

  • innodb_buffer_free:当前已分配给表的非数据部分使用的buffer pool字节总数(即空闲页所在的字节数,计算公式:innodb_buffer_allocated - innodb_buffer_data)

  • innodb_buffer_pages:当前已分配给表的buffer pool总页数

  • innodb_buffer_pages_hashed:当前已分配给表的自适应hash索引页总数

  • innodb_buffer_pages_old:当前已分配给表的旧页总数(位于LRU列表中的旧块子列表中的页数)

  • innodb_buffer_rows_cached:buffer pool中为表缓冲的总数据行数

07.schema_unused_indexes

查看不活跃的索引(没有任何事件发生的索引,这表示该索引从未使用过),默认情况下按照schema名称和表名进行排序。数据来源:performance_schema.table_io_waits_summary_by_index_usage

  • 该视图在server启动之后运行足够长的时间之后,所查询出的数据才比较适用,否则该视图查询的数据可能并不十分可靠,因为统计的数据可能并不精确,有一部分业务查询逻辑可能还来不及查询

视图查询语句文本

SELECT object_schema,
  object_name,
  index_name
FROM performance_schema.table_io_waits_summary_by_index_usage
WHERE index_name IS NOT NULL
AND count_star = 0
AND object_schema != 'mysql'
AND index_name != 'PRIMARY'
ORDER BY object_schema, object_name;

下面我们看看使用该视图查询返回的结果

admin@localhost : sys 12:40:28> select * from schema_unused_indexes limit 3;
+---------------+-------------+-------------------+
| object_schema | object_name | index_name        |
+---------------+-------------+-------------------+
| luoxiaobo    | public_num  | public_name_index |
| sbtest        | sbtest1    | k_1              |
| sbtest        | sbtest2    | k_2              |
+---------------+-------------+-------------------+
3 rows in set (0.00 sec)

视图字段含义如下:

  • object_schema:schema名称

  • OBJECT_NAME:表名

  • INDEX_NAME:未使用的索引名称

本期内容就介绍到这里,本期内容参考链接如下:

https://dev.mysql.com/doc/refman/5.7/en/sys-schema-unused-indexes.html

https://dev.mysql.com/doc/refman/5.7/en/sys-schema-auto-increment-columns.html

https://dev.mysql.com/doc/refman/5.7/en/sys-schema-index-statistics.html

https://dev.mysql.com/doc/refman/5.7/en/sys-schema-object-overview.html

https://dev.mysql.com/doc/refman/5.7/en/sys-schema-redundant-indexes.html

https://dev.mysql.com/doc/refman/5.7/en/sys-schema-table-statistics.html

https://dev.mysql.com/doc/refman/5.7/en/sys-schema-table-statistics-with-buffer.html

| 作者简介

罗小波·沃趣科技高级数据库技术专家

IT从业多年,历任运维工程师,高级运维工程师,运维经理,数据库工程师,曾参与版本发布系统,轻量级监控系统,运维管理平台,数据库管理平台的设计与编写,熟悉MySQL的体系结构时,InnoDB存储引擎,喜好专研开源技术,追求完美。

发表评论

电子邮件地址不会被公开。 必填项已用*标注