mysql innodb空间_【Mysql】innodb_space 的使用介绍

1、安装

# yum -y install ruby

# gem install innodb_ruby

安装完成后，执行如下命令验证innodb_ruby是否安装成功：# innodb_space --help

2、mysql环境

innodb_file_per_table=ON，innodb_file_format=Barracuda, innodb_file_format_max=Barracuda

3.基本用法

Against a single space file (ibdata or .ibd):

Option Parameters Description

-f Load the tablespace file (system or table)

Against a system tablespace which will auto-load file-per-table tablespace files:

Option Parameters Description

-s Load the system tablespace file (e.g. ibdata1)

-T

-I Use the given index name

4.常用的用法

虽然每个innodb表都有frm文件，不过innodb在系统表空间中ibdata1也维护着innodb表的元数据，所以直接分析ibdata1文件即可了解某个表的相关存储信息

表结构：

CREATE TABLE `aa` (

`id` int(11) unsigned NOT NULL AUTO_INCREMENT COMMENT '地区ID',

`name` varchar(20) NOT NULL DEFAULT '' COMMENT '名称',

`rel_id` varchar(50) NOT NULL DEFAULT '' COMMENT '关系ID',

`pid` int(10) unsigned NOT NULL DEFAULT '0' COMMENT '父ID',

`level` int(11) NOT NULL DEFAULT '0' COMMENT '类别，1、省份 2、市 3、区 4、县',

PRIMARY KEY (`id`),

UNIQUE KEY `UNQ_RID` (`rel_id`) USING BTREE,

KEY `IDX_PID` (`pid`)

) ENGINE=InnoDB AUTO_INCREMENT=1 DEFAULT CHARSET=utf8 COMMENT='地区表'

1 row in set (0.00 sec)

表信息：

4.1 space 相关

4.1.1 system-spaces

innodb_space -s ibdata1 system-spaces //列出系统中可用的所有表空间，包括一些基本的统计信息。这基本上是一个表的列表:

innodb_space -s ibdata1 system-spaces

[root@vm-test01 ztj]# innodb_space -s ../ibdata1 system-spaces

name pages indexes

(system) 4864 6

monitor/monitor 6 1

monitor/db_monitor 6 1

monitor/monitor 6 1

mysql/innodb_index_stats 6 1

mysql/innodb_table_stats 6 1

mysql/slave_master_info 6 1

mysql/slave_relay_log_info 6 1

mysql/slave_worker_info 6 1

opsdb/myapp_db_instance 8 1

test/aaa 6 1

test/area 8 3

ztj/aa 9 3

ztj/aaa 8 3

ztj/dim_library 4608 4

4.1.2 space-page-type-regions

//查看数据分布

innodb_space -s ibdata1 -T ztj/aa space-page-type-regions

[root@vm-test01 data]# innodb_space -s ibdata1 -T ztj/aa space-page-type-regions

start end count type

0 0 1 FSP_HDR

1 1 1 IBUF_BITMAP

2 2 1 INODE

3 5 3 INDEX //索引分布在page：3/4/5，其中3保存Primary信息，4保存UNQ_RID信息，5保存IDX_PID信息

6 7 2 FREE (ALLOCATED)

对于表的索引的名称，也可以通过information_schema库表innodb_sys_indexes、innodb_sys_tables查看：

mysql> SELECT

-> b.name, a.name, index_id, type, a.space, a.PAGE_NO

-> FROM

-> information_schema.INNODB_SYS_INDEXES a,

-> information_schema.INNODB_SYS_TABLES b

-> WHERE

-> a.table_id = b.table_id AND a.space <> 0;

+------------------------------+-------------------------------------------+----------+------+-------+---------+

| name | name | index_id | type | space | PAGE_NO |

+------------------------------+-------------------------------------------+----------+------+-------+---------+

| monitor/ll_monitor | GEN_CLUST_INDEX | 21 | 1 | 5 | 3 |

| monitor/db_monitor | GEN_CLUST_INDEX | 22 | 1 | 6 | 3 |

| monitor/monitor | GEN_CLUST_INDEX | 20 | 1 | 4 | 3 |

| mysql/innodb_index_stats | PRIMARY | 23 | 3 | 7 | 3 |

| mysql/innodb_table_stats | PRIMARY | 24 | 3 | 8 | 3 |

| mysql/slave_master_info | PRIMARY | 25 | 3 | 9 | 3 |

| mysql/slave_relay_log_info | PRIMARY | 26 | 3 | 10 | 3 |

| mysql/slave_worker_info | PRIMARY | 47 | 3 | 18 | 3 |

| ztj/aa | UNQ_RID | 51 | 2 | 19 | 4 |

| ztj/aa | IDX_PID | 62 | 0 | 19 | 6 |

| ztj/aaa | PRIMARY | 63 | 3 | 25 | 3 |

| ztj/aaa | UNQ_RID | 64 | 2 | 25 | 4 |

| ztj/aaa | IDX_PID | 65 | 0 | 25 | 5 |

+------------------------------+-------------------------------------------+----------+------+-------+---------+

4.1.3 space-indexes

//索引结构、数据分配情况

innodb_space -s ibdata1 -T ztj/aa space-indexes

[root@vm-test01 data]# innodb_space -s ibdata1 -T ztj/aa space-indexes

id name root fseg used allocated fill_factor

50 PRIMARY 3 internal 1 1 100.00%

50 PRIMARY 3 leaf 0 0 0.00%

51 UNQ_RID 4 internal 1 1 100.00%

51 UNQ_RID 4 leaf 0 0 0.00%

52 IDX_PID 5 internal 1 1 100.00%

52 IDX_PID 5 leaf 0 0 0.00%

4.1.4 space-page-type-summary

//查看表pages的占比

innodb_space -s ../ibdata1 -T ztj/aa space-page-type-summary

[root@vm-test01 ztj]# innodb_space -s ../ibdata1 -T ztj/aa space-page-type-summary

type count percent description

INDEX 3 37.50 B+Tree index

ALLOCATED 2 25.00 Freshly allocated

INODE 1 12.50 File segment inode

IBUF_BITMAP 1 12.50 Insert buffer bitmap

FSP_HDR 1 12.50 File space header

4.1.5 space-index-pages-summary

//查看表的索引分布在pages的情况

innodb_space -s ibdata1 -T ztj/aaa space-index-pages-summar |head -n 10

innodb_space -f ztj/aaa.ibd space-index-pages-summary |head -n 10

[root@vm-test01 data]# innodb_space -s ibdata1 -T ztj/aaa space-index-pages-summary |head -n 10

page index level data free records

3 63 0 948 15298 16 //主键B+树的高度为page level+1=1

4 64 0 380 15866 16

5 65 0 208 16040 16

6 0 0 0 16384 0

7 0 0 0 16384 0

[root@vm-test01 ztj]# innodb_space -f dim_library.ibd space-index-pages-summary |head -n 10

page index level data free records

3 28 2 45 16207 3 //主键B+树索引的高度为page level+1=3

4 29 1 8470 7598 385

5 30 1 11487 4575 547

6 31 1 7980 8122 380

7 28 0 7526 8690 72

8 28 0 15137 1045 144

9 28 0 15074 1108 144

10 28 0 15110 1072 144

11 28 0 15104 1080 143

4.2 page 相关

4.2.1 page-account

//查看该page的说明信息

[root@vm-test01 ztj]# innodb_space -s ../ibdata1 -T ztj/aa -p 3 page-account

Accounting for page 3:

Page type is INDEX (B+Tree index, table and index data stored in B+Tree structure).

Extent descriptor for pages 0-63 is at page 0, offset 158.

Extent is not fully allocated to an fseg; may be a fragment extent.

Page is marked as used in extent descriptor.

Extent is in free_frag list of space.

Page is in fragment array of fseg 1.

Fseg is in internal fseg of index 63.

Index root is page 3.

Index is ztj /aa.PRIMARY.

[root@vm-test01 ztj]#

[root@vm-test01 ztj]# innodb_space -s ../ibdata1 -T ztj/aa -p 4 page-account

Accounting for page 4:

Page type is INDEX (B+Tree index, table and index data stored in B+Tree structure).

Extent descriptor for pages 0-63 is at page 0, offset 158.

Extent is not fully allocated to an fseg; may be a fragment extent.

Page is marked as used in extent descriptor.

Extent is in free_frag list of space.

Page is in fragment array of fseg 3.

Fseg is in internal fseg of index 64.

Index root is page 4.

Index is ztj/aa.UNQ_RID.

4.2.2 page-records

//查看辅助索引存储信息

innodb_space -s ibdata1 -T ztj/aa -p 4 page-records

[root@vm-test01 data]# innodb_space -s ibdata1 -T ztj/aa -p 4 page-records

Record 126: (rel_id="10000") → (id=10000)

Record 141: (rel_id="1000010001") → (id=10001)

Record 161: (rel_id="100001000110002") → (id=10002)

Record 186: (rel_id="100001000110003") → (id=10003)

Record 211: (rel_id="100001000110004") → (id=10004)

Record 236: (rel_id="100001000110005") → (id=10005)

Record 261: (rel_id="100001000110006") → (id=10006)

Record 286: (rel_id="100001000110007") → (id=10007)

Record 311: (rel_id="100001000110008") → (id=10008)

Record 336: (rel_id="1000010009") → (id=10009)

Record 356: (rel_id="100001000910010") → (id=10010)

Record 381: (rel_id="100001000910011") → (id=10011)

Record 406: (rel_id="100001000910012") → (id=10012)

Record 431: (rel_id="100001000910013") → (id=10013)

Record 456: (rel_id="100001000910014") → (id=10014)

Record 481: (rel_id="100001000910015") → (id=10015)

innodb_space -s ibdata1 -T ztj/aa -p 5 page-records

[root@vm-test01 data]# innodb_space -s ibdata1 -T ztj/aa -p 5 page-records

Record 125: (pid=0) → (id=10000)

Record 138: (pid=10000) → (id=10001)

Record 151: (pid=10000) → (id=10009)

Record 164: (pid=10001) → (id=10002)

Record 177: (pid=10001) → (id=10003)

Record 190: (pid=10001) → (id=10004)

Record 203: (pid=10001) → (id=10005)

Record 216: (pid=10001) → (id=10006)

Record 229: (pid=10001) → (id=10007)

Record 242: (pid=10001) → (id=10008)

Record 255: (pid=10009) → (id=10010)

Record 268: (pid=10009) → (id=10011)

Record 281: (pid=10009) → (id=10012)

Record 294: (pid=10009) → (id=10013)

Record 307: (pid=10009) → (id=10014)

Record 320: (pid=10009) → (id=10015)

4.2.3 page-records

//主键索引的存储信息

innodb_space -s ibdata1 -T ztj/aa -p 3 page-records

[root@vm-test01 data]# innodb_space -s ibdata1 -T ztj/aa -p 3 page-records

Record 127: (id=10000) → (name="\xE5\x8C\x97\xE4\xBA\xAC", rel_id="10000", pid=0, level=1)

Record 170: (id=10001) → (name="\xE6\x9C\x9D\xE9\x98\xB3\xE5\x8C\xBA", rel_id="1000010001", pid=10000, level=2)

Record 221: (id=10002) → (name="\xE4\xB8\x89\xE7\x8E\xAF\xE4\xBB\xA5\xE5\x86\x85", rel_id="100001000110002", pid=10001, level=3)

Record 280: (id=10003) → (name="\xE4\xB8\x89\xE7\x8E\xAF\xE5\x88\xB0\xE5\x9B\x9B\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000110003", pid=10001, level=3)

Record 348: (id=10004) → (name="\xE5\x9B\x9B\xE7\x8E\xAF\xE5\x88\xB0\xE4\xBA\x94\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000110004", pid=10001, level=3)

Record 416: (id=10005) → (name="\xE4\xBA\x94\xE7\x8E\xAF\xE5\x88\xB0\xE5\x85\xAD\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000110005", pid=10001, level=3)

Record 484: (id=10006) → (name="\xE7\xAE\xA1\xE5\xBA\x84", rel_id="100001000110006", pid=10001, level=3)

Record 537: (id=10007) → (name="\xE5\x8C\x97\xE8\x8B\x91", rel_id="100001000110007", pid=10001, level=3)

Record 590: (id=10008) → (name="\xE5\xAE\x9A\xE7\xA6\x8F\xE5\xBA\x84", rel_id="100001000110008", pid=10001, level=3)

Record 646: (id=10009) → (name="\xE6\xB5\xB7\xE6\xB7\x80\xE5\x8C\xBA", rel_id="1000010009", pid=10000, level=2)

Record 697: (id=10010) → (name="\xE4\xB8\x89\xE7\x8E\xAF\xE4\xBB\xA5\xE5\x86\x85", rel_id="100001000910010", pid=10009, level=3)

Record 756: (id=10011) → (name="\xE4\xB8\x89\xE7\x8E\xAF\xE5\x88\xB0\xE5\x9B\x9B\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000910011", pid=10009, level=3)

Record 824: (id=10012) → (name="\xE5\x9B\x9B\xE7\x8E\xAF\xE5\x88\xB0\xE4\xBA\x94\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000910012", pid=10009, level=3)

Record 892: (id=10013) → (name="\xE4\xBA\x94\xE7\x8E\xAF\xE5\x88\xB0\xE5\x85\xAD\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000910013", pid=10009, level=3)

Record 960: (id=10014) → (name="\xE5\x85\xAD\xE7\x8E\xAF\xE4\xBB\xA5\xE5\xA4\x96", rel_id="100001000910014", pid=10009, level=3)

Record 1019: (id=10015) → (name="\xE8\xA5\xBF\xE4\xB8\x89\xE6\x97\x97", rel_id="100001000910015", pid=10009, level=3)

结论：辅助索引存储了主键值。主键page存储了所有数据

4.3 index相关

4.3.1 index-recurse

//递归整个B+树，扫描所有页面

innodb_space -s ../ibdata1 -T ztj/aaa -I PRIMARY index-recurse

[root@vm-test01 ztj]# innodb_space -s ../ibdata1 -T ztj/aaa -I PRIMARY index-recurse

ROOT NODE #3: 16 records, 948 bytes

RECORD: (id=10000) → (name="\xE5\x8C\x97\xE4\xBA\xAC", rel_id="10000", pid=0, level=1)

RECORD: (id=10001) → (name="\xE6\x9C\x9D\xE9\x98\xB3\xE5\x8C\xBA", rel_id="1000010001", pid=10000, level=2)

RECORD: (id=10002) → (name="\xE4\xB8\x89\xE7\x8E\xAF\xE4\xBB\xA5\xE5\x86\x85", rel_id="100001000110002", pid=10001, level=3)

RECORD: (id=10003) → (name="\xE4\xB8\x89\xE7\x8E\xAF\xE5\x88\xB0\xE5\x9B\x9B\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000110003", pid=10001, level=3)

RECORD: (id=10004) → (name="\xE5\x9B\x9B\xE7\x8E\xAF\xE5\x88\xB0\xE4\xBA\x94\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000110004", pid=10001, level=3)

RECORD: (id=10005) → (name="\xE4\xBA\x94\xE7\x8E\xAF\xE5\x88\xB0\xE5\x85\xAD\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000110005", pid=10001, level=3)

RECORD: (id=10006) → (name="\xE7\xAE\xA1\xE5\xBA\x84", rel_id="100001000110006", pid=10001, level=3)

RECORD: (id=10007) → (name="\xE5\x8C\x97\xE8\x8B\x91", rel_id="100001000110007", pid=10001, level=3)

RECORD: (id=10008) → (name="\xE5\xAE\x9A\xE7\xA6\x8F\xE5\xBA\x84", rel_id="100001000110008", pid=10001, level=3)

RECORD: (id=10009) → (name="\xE6\xB5\xB7\xE6\xB7\x80\xE5\x8C\xBA", rel_id="1000010009", pid=10000, level=2)

RECORD: (id=10010) → (name="\xE4\xB8\x89\xE7\x8E\xAF\xE4\xBB\xA5\xE5\x86\x85", rel_id="100001000910010", pid=10009, level=3)

RECORD: (id=10011) → (name="\xE4\xB8\x89\xE7\x8E\xAF\xE5\x88\xB0\xE5\x9B\x9B\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000910011", pid=10009, level=3)

RECORD: (id=10012) → (name="\xE5\x9B\x9B\xE7\x8E\xAF\xE5\x88\xB0\xE4\xBA\x94\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000910012", pid=10009, level=3)

RECORD: (id=10013) → (name="\xE4\xBA\x94\xE7\x8E\xAF\xE5\x88\xB0\xE5\x85\xAD\xE7\x8E\xAF\xE4\xB9\x8B\xE9\x97\xB4", rel_id="100001000910013", pid=10009, level=3)

RECORD: (id=10014) → (name="\xE5\x85\xAD\xE7\x8E\xAF\xE4\xBB\xA5\xE5\xA4\x96", rel_id="100001000910014", pid=10009, level=3)

RECORD: (id=10015) → (name="\xE8\xA5\xBF\xE4\xB8\x89\xE6\x97\x97", rel_id="100001000910015", pid=10009, level=3)

4.3.2 index-record-offsets

//像index-record一样递归整个索引树，不过只打印出index page的偏移量

innodb_space -s ../ibdata1 -T ztj/aaa -I PRIMARY index-record-offsets

[root@vm-test01 ztj]# innodb_space -s ../ibdata1 -T ztj/aaa -I PRIMARY index-record-offsets

page_offset record_offset

3 127

3 170

3 221

3 280

3 348

3 416

3 484

3 537

3 590

3 646

3 697

3 756

3 824

3 892

3 960

3 1019

4.4 record相关

4.4.1 record-history

//递归整个B+树，扫描所有页面

innodb_space -s ../ibdata1 -T ztj/aaa -I PRIMARY index-recurse

[root@vm-test01 ztj]# innodb_space -s ../ibdata1 -T ztj/aaa -I PRIMARY -p 3 -R 127 record-history

Transaction Type Undo record

(n/a) insert (id=10000) → ()

5、索引是物理有序的还是逻辑有序的？

page内的数据：逻辑上有序的，物理磁盘上⼀个page内是⽆序的，以指针的形式连接

数据，达到逻辑有序 【正常1, 2, 3, 4, 5. 存储是：1， 3 ，2， 5， 4。1链接2链接3链

接4链接5】⻚和⻚之间也是：逻辑上有序，物理上⽆序。 ⾃增ID，逻辑上也是有序

的。