背景: 再一次看完MySQL 技术内幕-Innodb存储引擎 一书的的第4章。对前面五节的内容做又有了新的认识，顺便做下笔记。先了解下相关的概念： 表空间： INNODB 所有数据都存在表空间当中（共享表空间），要是开启innodb_file_per_table，则每张表的数据会存到单
背景:
再一次看完
表空间：INNODB 所有数据都存在表空间当中（共享表空间），要是开启innodb_file_per_table，则每张表的数据会存到单独的一个表空间内（独享表空间）。
独享表空间包括：数据，索引，插入缓存，数据字典。共享表空间包括：Undo信息（不会回收），双写缓存信息，事务信息等。
段（segment）：组成表空间，有区组成。
区（extent）：有64个连续的页组成。每个页16K，总共1M。对于大的数据段，每次最后可申请4个区。
页（page）：是INNODB 磁盘管理的单位，有行组成。
行（row）：包括事务ID，回滚指针，列信息等。
目的1：
了解表空间各个页的信息和溢出行数据存储的信息。通过该书作者蒋承尧编写的工具：http://code.google.com/p/david-mysql-tools/source/browse/trunk/py_innodb_page_type/
3个脚本：
py_innodb_page_info.py
<span>#</span><span>! /usr/bin/env python </span><span>
#</span><span>encoding=utf-8</span>
<span>import</span><span> mylib
</span><span>from</span> sys <span>import</span><span> argv
</span><span>from</span> mylib <span>import</span><span> myargv
</span><span>if</span> <span>__name__</span> == <span>'</span><span>__main__</span><span>'</span><span>:
myargv </span>=<span> myargv(argv)
</span><span>if</span> myargv.parse_cmdline() ==<span> 0:
</span><span>pass</span>
<span>else</span><span>:
mylib.get_innodb_page_type(myargv)</span>
mylib.py
View Code
include.py
View Code
测试1：
root<span>@localhost</span> : test <span>02</span>:<span>26</span>:<span>13</span><span>></span><span>create</span> <span>table</span> tt(id <span>int</span> auto_increment,name <span>varchar</span>(<span>10</span>),age <span>int</span>,address <span>varchar</span>(<span>20</span>),<span>primary</span> <span>key</span> (id))engine<span>=</span><span>innodb; Query OK, </span><span>0</span> rows affected (<span>0.17</span><span> sec) root</span><span>@zhoujy</span>:<span>/</span><span>var</span><span>/</span>lib<span>/</span>mysql<span>/</span>test# ls <span>-</span><span>lh tt.ibd </span><span>-</span>rw<span>-</span>rw<span>--</span><span>-- 1 mysql mysql 96K 2012-10-17 14:26 tt.ibd</span>
查看ibd：
root@zhoujy:/home/zhoujy/jiaoben/read_ibd<span>#</span><span> python py_innodb_page_info.py /var/lib/mysql/test/tt.ibd -v</span> page offset 00000000, page type <file space header><span> page offset </span>00000001, page type <insert buffer bitmap><span> page offset </span>00000002, page type <file segment inode><span> page offset </span>00000003, page type <b-tree node>, page level ---叶子节点<span> page offset </span>00000000, page type <freshly allocated page><span> page offset </span>00000000, page type <freshly allocated page><span> Total number of page: </span>6<span>: Freshly Allocated Page: </span>2<span> Insert Buffer Bitmap: </span>1<span> File Space Header: </span>1<span> B</span>-tree Node: 1<span> File Segment inode: </span>1</freshly></freshly></b-tree></file></insert></file>
解释：
Total number of page: 总页数
Freshly Allocated Page:可用页
Insert Buffer Bitmap:插入缓存位图页
Insert Buffer Free List:插入缓存空闲列表页
B-tree Node:数据页
Uncompressed BLOB Page:二进制大对象页，存放溢出行的页，即溢出页
上面得到的信息是表初始化大小为96K，他是有 Total number of page * 16 得来的。1个数据页，2个可用页面。
root<span>@localhost</span> : test <span>02</span>:<span>42</span>:<span>58</span><span>></span><span>insert</span> <span>into</span> tt <span>values</span>(name,age,address) <span>values</span>(<span>'</span><span>aaa</span><span>'</span>,<span>23</span>,<span>'</span><span>HZZZ</span><span>'</span>);
疑惑：为什么没有申请区？区是64个连续的页，大小1M。那么表大小也应该是至少1M。但是现在只有96K（默认）。原因是因为每个段开始的时候，先有32个页大小的碎片页存放数据，使用
完之后才是64页的连续申请，最多每次可以申请4个区，保证数据的顺序。这里看出表大小增加是按照至少64页的大小的空间来增加的，即1M增加。
验证：
填充数据，写满这32个碎片页，32*16 = 512K。看看是否能申请大于1M的空间。
View Code
"额外"页：4个
page offset 00000000, page type
page offset 00000001, page type
page offset 00000002, page type
page offset 00000003, page type
碎片页：32个
page type
总共36个页，ibd大小 576K的由来：32*16=512K（碎片页）+ 4*16=64（额外页），这里开始要是再插入的话，应该申请最少1M的页：
root@zhoujy:/home/zhoujy/jiaoben/read_ibd<span>#</span><span> ls -lh /var/lib/mysql/test/tt.ibd </span> -rw-rw---- 1 mysql mysql 2.0M 2012-10-17 16:10 /var/lib/mysql/test/<span>tt.ibd root</span>@zhoujy:/home/zhoujy/jiaoben/read_ibd<span>#</span><span> python py_innodb_page_info.py /var/lib/mysql/test/tt.ibd</span> Total number of page: 128<span>: Freshly Allocated Page: </span>91<span> Insert Buffer Bitmap: </span>1<span> File Space Header: </span>1<span> B</span>-tree Node: 34<span> File Segment inode: </span>1
页从36跳到了128，因为已经用完了32个碎片页，新的页会采用区的方式进行空间申请。信息中看到有很多可用页，正好说明这点。
▲溢出行数据存放：INNODB存储引擎是索引组织的，即每页中至少有两行记录，因此如果页中只能存放一行记录，INNODB会自动将行数据放到溢出页中。当发生溢出行的时候，实际数据保存在BLOB页中，数据页只保存数据的前768字节（老的文件格式），新的文件格式（Barracuda）采用完全行溢出的方式，数据页只保存20个字节的指针，BLOB也保存所有数据。如何查看表中有溢出行数据呢？
root<span>@localhost</span> : test <span>04</span>:<span>52</span>:<span>34</span><span>></span><span>create</span> <span>table</span> t1 (id <span>int</span>,name <span>varchar</span>(<span>10</span>),memo <span>varchar</span>(<span>8000</span>))engine <span>=</span>innodb <span>default</span><span> charset utf8; Query OK, </span><span>0</span> rows affected (<span>0.16</span><span> sec) root</span><span>@localhost</span> : test <span>04</span>:<span>53</span>:<span>10</span><span>></span><span>insert</span> <span>into</span> t1 <span>values</span>(<span>1</span>,<span>'</span><span>zjy</span><span>'</span>,repeat(<span>'</span><span>我</span><span>'</span>,<span>8000</span><span>)); Query OK, </span><span>1</span> row affected (<span>0.00</span> sec)
查看ibd：
root@zhoujy:/home/zhoujy/jiaoben/read_ibd<span>#</span><span> python py_innodb_page_info.py /var/lib/mysql/test/t1.ibd -v</span> page offset 00000000, page type <file space header><span> page offset </span>00000001, page type <insert buffer bitmap><span> page offset </span>00000002, page type <file segment inode><span> page offset </span>00000003, page type <b-tree node>, page level <span> page offset </span>00000004, page type <uncompressed blob page><span> page offset </span>00000005, page type <uncompressed blob page><span> Total number of page: </span>6<span>: Insert Buffer Bitmap: </span>1<span> Uncompressed BLOB Page: </span>2<span> File Space Header: </span>1<span> B</span>-tree Node: 1<span> File Segment inode: </span>1</uncompressed></uncompressed></b-tree></file></insert></file>
从信息中看到，刚才插入的一行记录，已经溢出了，保存到了2个BLOB页中（
root<span>@localhost</span> : test <span>05</span>:<span>08</span>:<span>39</span><span>></span><span>create</span> <span>table</span> t2 (id <span>int</span>,name <span>varchar</span>(<span>1000</span>),address <span>varchar</span>(<span>512</span>),company <span>varchar</span>(<span>200</span>),xx <span>varchar</span>(<span>512</span>),memo <span>varchar</span>(<span>512</span>),dem <span>varchar</span>(<span>1000</span>))engine <span>=</span>innodb <span>default</span><span> charset utf8; Query OK, </span><span>0</span> rows affected (<span>0.17</span><span> sec) root</span><span>@localhost</span> : test <span>05</span>:<span>08</span>:<span>43</span><span>></span><span>insert</span> <span>into</span> t2 <span>values</span>(<span>1</span>,repeat(<span>'</span><span>周</span><span>'</span>,<span>1000</span>),repeat(<span>'</span><span>我</span><span>'</span>,<span>500</span>),repeat(<span>'</span><span>丁</span><span>'</span>,<span>500</span>),repeat(<span>'</span><span>啊</span><span>'</span>,<span>500</span>),repeat(<span>'</span><span>噢</span><span>'</span>,<span>500</span>),repeat(<span>'</span><span>阿a</span><span>'</span>,<span>500</span>));
1000+500+500+500+500+500=3500*3>8000字节；行会被溢出：
root@zhoujy:/home/zhoujy/jiaoben/read_ibd<span>#</span><span> python py_innodb_page_info.py /var/lib/mysql/test/t2.ibd -v</span> page offset 00000000, page type <file space header><span> page offset </span>00000001, page type <insert buffer bitmap><span> page offset </span>00000002, page type <file segment inode><span> page offset </span>00000003, page type <b-tree node>, page level <span> page offset </span>00000004, page type <uncompressed blob page><span> page offset </span>00000000, page type <freshly allocated page><span> Total number of page: </span>6<span>: Insert Buffer Bitmap: </span>1<span> Freshly Allocated Page: </span>1<span> File Segment inode: </span>1<span> B</span>-tree Node: 1<span> File Space Header: </span>1<span> Uncompressed BLOB Page: </span>1</freshly></uncompressed></b-tree></file></insert></file>
root@zhoujy:/home/zhoujy/jiaoben/read_ibd<span>#</span><span> hexdump -C -v /var/lib/mysql/test/t1.ibd > t1.txt</span>
查看ibd：
View Code
文本中刚好是48行，每行16字节。48*16=768字节，刚好验证了之前说的：数据页只保存数据的前768字节（老的文件格式）。
总结1:
通过上面的信息，可以能清楚的知道ibd表空间各个页的分布和利用信息以及表空间大小增加的步长；特别注意的是溢出行，一个页中至少包含2行数据，如果页中存放的行数越多，性能就越好。
************************************
************************************
目的2：
了解表空间如何存储数据，以及对NULL值的存储。
测试2：
在测试前先了解INNODB的存储格式（row_format）。老格式（Antelope）：Compact，Redumdant；新格式（Barracuda）：Compressed ，Dynamic。
这里测试指针对默认的存储格式。
Compact行记录方式如下:
<span><strong> |变长字段长度列表（1~2字节）|NULL标志位（1字节）|记录头信息（5字节）|RowID（6字节）|事务ID（6字节）|回滚指针（7字节）|</strong></span>
上面信息除了 "NULL标志位"[表中所有字段都定义为NOT NULL]，"RowID"[表中有主键] ，"变长字段长度列表"
[没有变长字段] 可能不存在外，其他信息都会出现。所以一行数据除了列数据所占用的字段外，还需要额外18字节。
一:字段全NULL
mysql<span>></span> <span>create</span> <span>table</span> mytest(t1 <span>varchar</span>(<span>10</span>),t2 <span>varchar</span>(<span>10</span>),t3 <span>varchar</span>(<span>10</span>) ,t4 <span>varchar</span>(<span>10</span>))engine<span>=</span>innodb charset <span>=</span> latin1 row_format<span>=</span><span>compact; Query OK, </span><span>0</span> rows affected (<span>0.08</span><span> sec) mysql</span><span>></span> <span>insert</span> <span>into</span> mytest <span>values</span>(<span>'</span><span>a</span><span>'</span>,<span>'</span><span>bb</span><span>'</span>,<span>'</span><span>bb</span><span>'</span>,<span>'</span><span>ccc</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.02</span><span> sec) mysql</span><span>></span> <span>insert</span> <span>into</span> mytest <span>values</span>(<span>'</span><span>a</span><span>'</span>,<span>'</span><span>ee</span><span>'</span>,<span>'</span><span>ee</span><span>'</span>,<span>'</span><span>fff</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.01</span><span> sec) mysql</span><span>></span> <span>insert</span> <span>into</span> mytest <span>values</span>(<span>'</span><span>a</span><span>'</span>,<span>NULL</span>,<span>NULL</span>,<span>'</span><span>fff</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.00</span> sec)
测试数据准备完之后，执行shell命令：
root@zhoujy:/usr/local/mysql/test<span>#</span><span> hexdump -C -v mytest.ibd > /home/zhoujy/mytest.txt</span>
打开mytest.txt文件找到supremum这一行：
0000c070 73 75 70 72 65 6d 75 6d <span>03 02 02 01</span> <span>00</span> <span>00 00 10</span> |supremum........| -----------><strong>一行，16字节</strong><span> 0000c080 </span><span>00 25</span> <span>00 00 00 03 b9 00</span> <span>00 00 00 02 49 01</span> <span>82 00</span> |.%..........I...|<span> 0000c090 </span><span>00 01 4a 01 10</span> <span>61 62 62 62 62 63 63 63</span> 03 02 02 |..J..abbbbccc...|<span> 0000c0a0 </span>01 00 00 00 18 00 23 00 00 00 03 b9 01 00 00 00 |......<span>#</span><span>.........|</span> 0000c0b0 02 49 02 83 00 00 01 4b 01 10 61 65 65 65 65 66 |.I.....K..aeeeef|<span> 0000c0c0 </span>66 66 <span>03 01</span> <span>06</span> <span>00 00 20 ff a6</span> <span>00 00 00 03 b9 02</span> |ff..... ........|<span> 0000c0d0 </span><span>00 00 00 02 49 03</span> <span>84 00 00 01 4c 01 10</span> <span>61 66 66</span> |....I.....L..aff|<span> 0000c0e0 </span><span>66</span> 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |f...............|
解释：
第一行数据:
03 02 02 01 /*变长字段*/ ---- 表中4个字段类型为varchar，并且没有NULL数据，而且每个字段君小于255。
00 /*NULL标志位，第一行没有null的数据*/
00 00 10 00 25 /*记录头信息,固定5个字节*/
00 00 00 03 b9 00 /*RowID,固定6个字节,表没有主键*/
00 00 00 02 49 01 /*事务ID,固定6个字节*/
82 00 00 01 4a 01 10 /*回滚指针,固定7个字节*/
61 62 62 62 62 63 63 63 /*列的数据*/
第二行数据和第一行数据一样（颜色匹配）。
第三行数据（有NULL值）和第一行的解释的颜色对应起来比较差别:
<span><span>03 02 02 01</span> VS </span><span><span>03 01</span><span><span> ----------</span><span><span>当值为NULL时，变长字段列表<span><span>不会占用</span></span>存储空间。</span></span></span></span>
<span><span>61 62 62 62 62 63 63 63</span> VS </span><span><span>61 66 66 </span><span><span>66</span><span><span> ---------</span> <span>NULL值没有存储，不占空间</span></span></span></span>
结论：当值为NULL时，变长字段列表不会占用存储空间。NULL值没有存储，不占空间，但是需要一个标志位（一行一个）。
二:字段全NOT NULL
mysql<span>></span> <span>create</span> <span>table</span> mytest(t1 <span>varchar</span>(<span>10</span>) <span>NOT</span> <span>NULL</span>,t2 <span>varchar</span>(<span>10</span>) <span>NOT</span> <span>NULL</span>,t3 <span>varchar</span>(<span>10</span>) <span>NOT</span> <span>NULL</span>,t4 <span>varchar</span>(<span>10</span>) <span>NOT</span> <span>NULL</span>)engine<span>=</span>innodb charset <span>=</span> latin1 row_format<span>=</span><span>compact; Query OK, </span><span>0</span> rows affected (<span>0.03</span><span> sec) mysql</span><span>></span> <span>insert</span> <span>into</span> mytest <span>values</span>(<span>'</span><span>a</span><span>'</span>,<span>'</span><span>bb</span><span>'</span>,<span>'</span><span>bb</span><span>'</span>,<span>'</span><span>ccc</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.01</span><span> sec) mysql</span><span>></span> <span>insert</span> <span>into</span> mytest <span>values</span>(<span>'</span><span>a</span><span>'</span>,<span>'</span><span>ee</span><span>'</span>,<span>'</span><span>ee</span><span>'</span>,<span>'</span><span>fff</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.01</span><span> sec) mysql</span><span>></span> <span>insert</span> <span>into</span> mytest <span>values</span>(<span>'</span><span>a</span><span>'</span>,<span>NULL</span>,<span>NULL</span>,<span>'</span><span>fff</span><span>'</span><span>); ERROR </span><span>1048</span> (<span>23000</span>): <span>Column</span> <span>'</span><span>t2</span><span>'</span> cannot be <span>null</span>
步骤和上面一样，得到的ibd的结果是：
<span>0000c070 73 75 70 72 65 6d 75 6d <span>03 02 02 01</span> <span>00 00 10 00</span> |supremum........| 0000c080 <span>24</span> <span>00 00 00 03 b9 03</span> <span>00 00 00 02 49 07</span> <span>87 00 00</span> |$..........I....| 0000c090 <span>01 4f 01 10</span> <span>61 62 62 62 62 63 63 63</span> 03 02 02 01 |.O..abbbbccc....| 0000c0a0 00 00 18 ff cb 00 00 00 03 b9 04 00 00 00 02 49 |...............I| 0000c0b0 08 88 00 00 01 50 01 10 61 65 65 65 65 66 66 66 |.....P..aeeeefff|</span>
和上面比较,发现少了NULL的标志位信息。
结论: NULL值会有额外的空间来存储，即每行1字节的大小。对于相同数据的表，字段中有NULL值的表比NOT NULL的大。
三:1个NULL，和1个''的数据:
mysql<span>></span> <span>create</span> <span>table</span> mytest(t1 <span>varchar</span>(<span>10</span>) <span>NOT</span> <span>NULL</span>,t2 <span>varchar</span>(<span>10</span>) <span>NOT</span> <span>NULL</span> <span>DEFAULT</span> <span>''</span>,t3 <span>varchar</span>(<span>10</span>) <span>NOT</span> <span>NULL</span> ,t4 <span>varchar</span>(<span>10</span>))engine<span>=</span>innodb charset <span>=</span> latin1 row_format<span>=</span><span>compact; Query OK, </span><span>0</span> rows affected (<span>0.02</span><span> sec) mysql</span><span>></span> <span>insert</span> <span>into</span> mytest(t1,t2) <span>values</span>(<span>'</span><span>A</span><span>'</span>,<span>'</span><span>BB</span><span>'</span><span>); Query OK, </span><span>1</span> row affected, <span>1</span> warning (<span>0.01</span> sec)
步骤和上面一样，得到的ibd的结果是：
0000c070 73 75 70 72 65 6d 75 6d <span>00 02 01</span> <span>01</span> <span>00 00 10 ff</span> |supremum........|<span> 0000c080 <span>ef </span></span><span>00 00 00 43 b9 03</span> <span>00 00 00 02 4a 15</span> <span>90 00 00</span> |....C......J....|<span> 0000c090 </span><span>01 c2 01 10</span> <span>41 42 42</span> 00 00 00 00 00 00 00 00 00 |....ABB.........|
和上面2个区别主要在于变长列表和列数据这里。
结论：列数据信息里表明了 NULL数据和''数据都不占用任何空间，对于变长字段列表的信息，和一对比得出：‘’数据虽然不需要占用任何存储空间，但是在变长字段列表里面还是需要占用一个字节，NULL值不需要占用”，只是NULL会有额外的一个标志位，所以能有个优化的说法：“数据库表中能设置NOT NULL的就尽量设置为NOT NULL，除非确实需要NULL值得。” 在此得到了证明。
上面的测试都是针对VARCHAR的变长类型，那对于CHAR呢？
CHAR 测试：
root<span>@localhost</span> : test <span>10</span>:<span>33</span>:<span>35</span><span>></span><span>create</span> <span>table</span> mytest(t1 <span>char</span>(<span>10</span>),t2 <span>char</span>(<span>10</span>),t3 <span>char</span>(<span>10</span>) ,t4 <span>char</span>(<span>10</span>))engine<span>=</span>innodb charset <span>=</span> latin1 row_format<span>=</span>compact;Query OK, <span>0</span> rows affected (<span>0.16</span><span> sec) root</span><span>@localhost</span> : test <span>10</span>:<span>33</span>:<span>59</span><span>></span><span>insert</span> <span>into</span> mytest <span>values</span>(<span>'</span><span>a</span><span>'</span>,<span>'</span><span>bb</span><span>'</span>,<span>'</span><span>bb</span><span>'</span>,<span>'</span><span>ccc</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.00</span><span> sec) root</span><span>@localhost</span> : test <span>10</span>:<span>34</span>:<span>09</span><span>></span><span>insert</span> <span>into</span> mytest <span>values</span>(<span>'</span><span>a</span><span>'</span>,<span>'</span><span>ee</span><span>'</span>,<span>'</span><span>ee</span><span>'</span>,<span>'</span><span>fff</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.00</span><span> sec) root</span><span>@localhost</span> : test <span>10</span>:<span>34</span>:<span>19</span><span>></span><span>insert</span> <span>into</span> mytest <span>values</span>(<span>'</span><span>a</span><span>'</span>,<span>NULL</span>,<span>NULL</span>,<span>'</span><span>fff</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.00</span> sec)
打开ibd生成的文件：
0000c060 02 00 1b 69 6e 66 69 6d 75 6d 00 04 00 0b 00 00 |...infimum......|<span> 0000c070 </span>73 75 70 72 65 6d 75 6d <span>00</span> <span>00 00 10 00 41</span> <span>00 00</span> |supremum.....A..|<span> 0000c080 </span><span>00 0a f5 00</span> <span>00 00 00 81 2d 07</span> <span>80 00 00 00 32 01</span> |........-.....2.|<span> 0000c090 </span><span>10</span> <span>61 20 20 20 20 20 20 20 20 20</span> <span>62 62 20 20 20</span> |.a bb |<span> 0000c0a0 </span><span>20 20 20 20 20</span> <span>62 62 20 20 20 20 20 20 20 20</span> <span>63</span> | bb c|<span> 0000c0b0 </span><span>63 63 20 20 20 20 20 20 20</span> 00 00 00 18 00 41 00 |cc .....A.|<span> 0000c0c0 </span>00 00 0a f5 01 00 00 00 81 2d 08 80 00 00 00 32 |.........-.....2|<span> 0000c0d0 </span>01 10 61 20 20 20 20 20 20 20 20 20 65 65 20 20 |..a ee |<span> 0000c0e0 </span>20 20 20 20 20 20 65 65 20 20 20 20 20 20 20 20 | ee |<span> 0000c0f0 </span>66 66 66 20 20 20 20 20 20 20 <span>06</span> <span>00 00 20 ff 70</span> |fff ... .p|<span> 0000c100 </span><span>00 00 00 0a f5 02</span> <span>00 00 00 81 2d 09</span> <span>80 00 00 00</span> |..........-.....|<span> 0000c110 </span><span>32 01 10</span> <span>61 20 20 20 20 20 20 20 20 20</span> <span>66 66 66</span> |2..a fff|<span> 0000c120 </span><span>20 20 20 20 20 20 20 00</span> 00 00 00 00 00 00 00 00 | .........|
和一的varchar比较发现：少了变长字段列表，但是对于char来讲，需要固定长度来存储的，存不到固定长度，也会被填充满。如：20；并且NULL值也不需要占用存储空间。
混合（varchar,char）:
root<span>@localhost</span> : test <span>11</span>:<span>21</span>:<span>48</span><span>></span><span>create</span> <span>table</span> mytest(t1 <span>int</span>,t2 <span>char</span>(<span>10</span>),t3 <span>varchar</span>(<span>10</span>) ,t4 <span>char</span>(<span>10</span>))engine<span>=</span>innodb charset <span>=</span> latin1 row_format<span>=</span><span>compact; Query OK, </span><span>0</span> rows affected (<span>0.17</span><span> sec) root</span><span>@localhost</span> : test <span>11</span>:<span>21</span>:<span>50</span><span>></span><span>insert</span> <span>into</span> mytest <span>values</span>(<span>1</span>,<span>'</span><span>a</span><span>'</span>,<span>'</span><span>b</span><span>'</span>,<span>'</span><span>c</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.00</span><span> sec) root</span><span>@localhost</span> : test <span>11</span>:<span>22</span>:<span>06</span><span>></span><span>insert</span> <span>into</span> mytest <span>values</span>(<span>11</span>,<span>'</span><span>aa</span><span>'</span>,<span>'</span><span>bb</span><span>'</span>,<span>'</span><span>cc</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.00</span> sec)
从上面的表结构中看出：
1，变长字段列表长度：1
2，NULL标志位：1
3，记录头信息：5
4，RowID：6
5，事务ID：6
6，回滚指针：7
idb的信息：
0000c070 73 75 70 72 65 6d 75 6d <span>01</span> <span>00</span> <span>00 00 10 00 33</span> <span>00</span> |supremum......3.| <span> 0000c080 </span><span>00 00 0a f5 07</span> <span>00 00 00 81 2d 1a</span> <span>80 00 00 00 32</span> |.........-.....2|<span> 0000c090 </span><span>01 10</span> <span>80 00 00 01</span> <span>61 20 20 20 20 20 20 20 20 20</span> |......a |<span> 0000c0a0 </span><span>62</span> <span>63 20 20 20 20 20 20 20 20 20</span> 02 00 00 00 18 |bc .....|<span> 0000c0b0 ff be </span>00 00 00 0a f5 08 00 00 00 81 2d 1b 80 00 |............-...|<span> 0000c0c0 </span>00 00 32 01 10 <span>80 00 00 0b</span> 61 61 20 20 20 20 20 |..2......aa |<span> 0000c0d0 </span>20 20 20 62 62 63 63 20 20 20 20 20 20 20 20 00 | bbcc .|
从上信息得出和之前预料的一样：因为表中只有一个varchar字段，所以，变长列表长度就只有：01
特别注意的是：各个列数据存储的信息：t1字段为int 类型，占用4个字节的大小。第一行：80 00 00 01 就是表示 1 数字；第二行：80 00 00 0b
表示了11的数字。[select hex(11) == B ]，其他的和上面的例子一样。
上面都是latin1单字节字符集的说明，那对于多字节字符集的情况怎么样？
root<span>@localhost</span> : test <span>11</span>:<span>52</span>:<span>10</span><span>></span><span>create</span> <span>table</span> mytest(id <span>int</span> auto_increment,t2 <span>varchar</span>(<span>10</span>),t3 <span>varchar</span>(<span>10</span>) ,t4 <span>char</span>(<span>10</span>),<span>primary</span> <span>key</span>(id))engine<span>=</span>innodb charset <span>=</span> utf8 row_format<span>=</span><span>compact; Query OK, </span><span>0</span> rows affected (<span>0.17</span><span> sec) root</span><span>@localhost</span> : test <span>11</span>:<span>52</span>:<span>11</span><span>></span><span>insert</span> <span>into</span> mytest(t2,t3,t4) <span>values</span>(<span>'</span><span>bb</span><span>'</span>,<span>'</span><span>bb</span><span>'</span>,<span>'</span><span>ccc</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.00</span><span> sec) root</span><span>@localhost</span> : test <span>11</span>:<span>55</span>:<span>34</span><span>></span><span>insert</span> <span>into</span> mytest(t2,t3,t4) <span>values</span>(<span>'</span><span>我们</span><span>'</span>,<span>'</span><span>他们</span><span>'</span>,<span>'</span><span>我们的</span><span>'</span><span>); Query OK, </span><span>1</span> row affected (<span>0.00</span> sec)
ibd信息如下：
0000c070 73 75 70 72 65 6d 75 6d <span>0a 02 02</span> <span>00</span> <span>00 00 10 00</span> |supremum........|<span> 0000c080 </span><span>28</span> <span>80 00 00 01 00 00</span> <span>00 81 2d 27 80 00 00</span> <span>00 32</span> |(........-'....2|<span> 0000c090 </span><span>01 10</span> 62 62 62 62 63 63 63 20 20 20 20 20 20 20 |..bbbbccc |<span> 0000c0a0 0a </span>06 06 00 00 00 18 ff c7 80 00 00 02 00 00 00 |................|<span> 0000c0b0 </span>81 2d 28 80 00 00 <span>00 32 01 10</span> <span>e6 88 91 e4 bb ac</span> |.-(....2........|<span> 0000c0c0 <span>e4 bb </span></span><span>96 e4 bb ac</span> <span>e6 88 91 e4 bb ac e7 9a 84</span> 20 |............... |
因为表有了主键，所以ROWID（6字节）不见了。
特别注意的是：变长字段列表是3？表里面的varchar类型的列只有2个啊。经测试得出：在多字节字符集的条件下，char类型被当成可变长度的类型来处理，他们的行存储基本没有区别，所以这个就出现变长列表是3了，因为是utf8字符集，占用三个字节。所以一个汉字均占用了一个页中3个字节的空间(”我们“ ：e6
88 91 e4 bb ac)。
数据列的信息：
id列的1值，应该是 80 00 00 01，为什么这个显示00
32 01 10，而且所有的id都是00
32 01 10。测试发现，id为自增主键的时候，id的4个字节长度都是以00
32 01 10 表示。否则和前面一个例子里说的，用select HEX（X） 表示。
总结2:
上面的测试都是基于COMPACT存储格式的，不管是varchar还是char，NULL值是不需要占用存储空间的；特别需要注意的是Redumdant的记录头信息需要6个固定字节，而NULL值对于varchar来说是不需要占用存储空间，对于char来说将会占用最大值的字节数；在多字节字符集的条件下，CHAR和VARCHAR的行存储基本是没有区别的。
