登录
首页 >  数据库 >  MySQL

【StoneDB技术解析】验证相关数据包是否需要解压缩

来源:SegmentFault

时间:2023-02-22 18:48:17 349浏览 收藏

本篇文章给大家分享《【StoneDB技术解析】验证相关数据包是否需要解压缩》,覆盖了数据库的常见基础知识,其实一个语言的全部知识点一篇文章是不可能说完的,但希望通过这些问题,让读者对自己的掌握程度有一定的认识(B 数),从而弥补自己的不足,更好的掌握它。

在StoneDB中,数据包分为以下几类:

  1. 不相关的数据包:不满足查询条件的数据包。
  2. 相关的数据包:满足查询条件的数据包。
  3. 可疑的数据包:数据包中的数据部分满足查询条件,需要进一步解压缩数据包才能得到满足条件的数据行。

通过对数据包的划分,知识网格技术过滤掉不相关的数据包,读取相关的数据包和可疑的数据包。其中相关的数据包不需要解压缩,只读取元数据,不会发生IO,可疑的数据包需要解压缩,会发生IO。

1)创建表t_user
CREATE TABLE t_user(
  id INT NOT NULL AUTO_INCREMENT,
  first_name VARCHAR(10) NOT NULL,
  last_name VARCHAR(10) NOT NULL,
  sex VARCHAR(5) NOT NULL,
  score INT NOT NULL,
  copy_id INT NOT NULL,
  PRIMARY KEY (`id`),
  key idx_lastname(last_name)
) engine=STONEDB;

2)创建存储过程
DELIMITER //
    create PROCEDURE add_user(in num INT)
    BEGIN
        DECLARE rowid INT DEFAULT 0;
        DECLARE firstname CHAR(1);
        DECLARE name1 CHAR(1);
        DECLARE name2 CHAR(1);
        DECLARE lastname VARCHAR(3) DEFAULT '';
        DECLARE sex CHAR(1);
        DECLARE score CHAR(2);
        WHILE rowid 

1)验证读取相关数据包

SQL的语义逻辑是对字段 first_name 进行分组统计,在StoneDB中,元数据信息记录在元数据节点,如果能通过元数据节点读取到元数据,就不需要解压缩数据包,不发生IO。
在InnoDB中,表的统计信息记录在mysql.innodb_table_stats,优化器根据表和索引的统计信息,生成一个最优的执行计划,然后执行SQL。分别在InnoDB与StoneDB执行,通过SQL profile观察读取IO的情况。
注:为规避缓存的影响,每组测试前重启数据库实例。

InnoDB

mysql> set profiling=on;                                          
Query OK, 0 rows affected, 1 warning (0.00 sec)

mysql> select first_name,count(*) from t_user_innodb group by first_name;
+------------+----------+
| first_name | count(*) |
+------------+----------+
| 侯         |   476424 |
| 刘         |   475764 |
| 吴         |   475979 |
| 周         |   475891 |
| 孙         |   950444 |
| 彭         |   476632 |
| 徐         |   476219 |
| 李         |   475521 |
| 杨         |   476026 |
| 林         |   477289 |
| 柳         |   476250 |
| 江         |   476623 |
| 王         |   475119 |
| 赵         |   476529 |
| 邹         |   476852 |
| 郑         |   476379 |
| 钱         |   476829 |
| 阮         |   476336 |
| 陈         |   476746 |
| 高         |   476148 |
+------------+----------+
20 rows in set (8.62 sec)

mysql> show profiles;                                                    
+----------+------------+-------------------------------------------------------------------+
| Query_ID | Duration   | Query                                                             |
+----------+------------+-------------------------------------------------------------------+
|        1 | 8.61591075 | select first_name,count(*) from t_user_innodb group by first_name |
+----------+------------+-------------------------------------------------------------------+
1 row in set, 1 warning (0.00 sec)

mysql> show profile cpu,block io for query 1;                            
+----------------------+----------+----------+------------+--------------+---------------+
| Status               | Duration | CPU_user | CPU_system | Block_ops_in | Block_ops_out |
+----------------------+----------+----------+------------+--------------+---------------+
| starting             | 0.000149 | 0.000059 |   0.000083 |            0 |             0 |
| checking permissions | 0.000027 | 0.000011 |   0.000015 |            0 |             0 |
| Opening tables       | 0.048181 | 0.003919 |   0.007952 |          608 |             0 |
| init                 | 0.000036 | 0.000014 |   0.000021 |            0 |             0 |
| System lock          | 0.000022 | 0.000009 |   0.000013 |            0 |             0 |
| optimizing           | 0.000017 | 0.000007 |   0.000010 |            0 |             0 |
| statistics           | 0.000029 | 0.000012 |   0.000016 |            0 |             0 |
| preparing            | 0.000022 | 0.000009 |   0.000013 |            0 |             0 |
| Creating tmp table   | 0.000045 | 0.000019 |   0.000027 |            0 |             0 |
| Sorting result       | 0.000016 | 0.000007 |   0.000009 |            0 |             0 |
| executing            | 0.000014 | 0.000005 |   0.000008 |            0 |             0 |
| Sending data         | 8.566974 | 6.905969 |   0.772964 |       873888 |             0 |
| Creating sort index  | 0.000144 | 0.000164 |   0.000037 |           64 |             0 |
| end                  | 0.000014 | 0.000012 |   0.000003 |           32 |             0 |
| query end            | 0.000028 | 0.000038 |   0.000009 |            0 |             0 |
| removing tmp table   | 0.000019 | 0.000015 |   0.000003 |            0 |             0 |
| query end            | 0.000012 | 0.000010 |   0.000002 |            0 |             0 |
| closing tables       | 0.000031 | 0.000025 |   0.000006 |            0 |             0 |
| freeing items        | 0.000032 | 0.000027 |   0.000006 |            0 |             0 |
| logging slow query   | 0.000067 | 0.000054 |   0.000012 |            0 |             8 |
| cleaning up          | 0.000035 | 0.000028 |   0.000006 |            0 |             0 |
+----------------------+----------+----------+------------+--------------+---------------+
21 rows in set, 1 warning (0.00 sec)

从SQL profile可知,SQL在InnoDB执行的过程中,发生IO的阶段有Opening tables、Sending data、Creating sort index、end,其中Opening tables是每张表第一次加载都会经历的,可排除讨论。重点关注Sending data部分,它表示在执行器的任意阶段,通常是存储引擎层与Server层的IO交互过程。

StoneDB

mysql> set profiling=on;
Query OK, 0 rows affected, 1 warning (0.00 sec)

mysql> select first_name,count(*) from t_user group by first_name;
+------------+----------+
| first_name | count(*) |
+------------+----------+
| 赵         |   476529 |
| 徐         |   476219 |
| 王         |   475119 |
| 阮         |   476336 |
| 柳         |   476250 |
| 侯         |   476424 |
| 孙         |   950444 |
| 郑         |   476379 |
| 高         |   476148 |
| 林         |   477289 |
| 邹         |   476852 |
| 彭         |   476632 |
| 李         |   475521 |
| 吴         |   475979 |
| 刘         |   475764 |
| 钱         |   476829 |
| 周         |   475891 |
| 杨         |   476026 |
| 陈         |   476746 |
| 江         |   476623 |
+------------+----------+
20 rows in set (0.59 sec)

mysql> show profiles;
+----------+------------+------------------------------------------------------------+
| Query_ID | Duration   | Query                                                      |
+----------+------------+------------------------------------------------------------+
|        1 | 0.59069975 | select first_name,count(*) from t_user group by first_name |
+----------+------------+------------------------------------------------------------+
1 row in set, 1 warning (0.00 sec)

mysql> show profile cpu,block io for query 1;
+----------------------+----------+----------+------------+--------------+---------------+
| Status               | Duration | CPU_user | CPU_system | Block_ops_in | Block_ops_out |
+----------------------+----------+----------+------------+--------------+---------------+
| starting             | 0.000160 | 0.000066 |   0.000089 |            0 |             0 |
| checking permissions | 0.000027 | 0.000011 |   0.000015 |            0 |             0 |
| Opening tables       | 0.011405 | 0.003718 |   0.007688 |            0 |           240 |
| System lock          | 0.000385 | 0.000163 |   0.000222 |            0 |             0 |
| init                 | 0.000050 | 0.000021 |   0.000028 |            0 |             0 |
| optimizing           | 0.000143 | 0.000061 |   0.000082 |            0 |             0 |
| update multi-index   | 0.000052 | 0.000022 |   0.000030 |            0 |             0 |
| aggregation          | 0.578315 | 2.639504 |   0.981471 |            0 |             8 |
| query end            | 0.000069 | 0.000043 |   0.000026 |            0 |             0 |
| closing tables       | 0.000035 | 0.000021 |   0.000013 |            0 |             0 |
| freeing items        | 0.000034 | 0.000021 |   0.000013 |            0 |             0 |
| cleaning up          | 0.000027 | 0.000017 |   0.000010 |            0 |             0 |
+----------------------+----------+----------+------------+--------------+---------------+
12 rows in set, 1 warning (0.00 sec)

从SQL profile可知,SQL在StoneDB执行的过程中,只在Opening tables阶段发生IO。其它阶段没有发生IO,说明相关数据包是不需要解压缩的,通过元数据得到。

2)验证读取可疑数据包

SQL的语义逻辑是查询一行数据,StoneDB可以通过知识网格技术过滤掉不相关的数据包,由于只返回一行数据,最终只能找到可疑的数据包,然后解压缩可疑的数据包,最终得到这一行数据。InnoDB还是根据统计信息生成一个最优的执行计划去执行SQL。

InnoDB

mysql> set profiling=on;                              
Query OK, 0 rows affected, 1 warning (0.00 sec)

mysql> select count(*) from t_user_innodb where first_name='柳' and copy_id=9968888;
+----------+
| count(*) |
+----------+
|        1 |
+----------+
1 row in set (3.20 sec)

mysql> show profile cpu,block io for query 1;
+----------------------+----------+----------+------------+--------------+---------------+
| Status               | Duration | CPU_user | CPU_system | Block_ops_in | Block_ops_out |
+----------------------+----------+----------+------------+--------------+---------------+
| starting             | 0.000170 | 0.000072 |   0.000092 |            0 |             0 |
| checking permissions | 0.000030 | 0.000013 |   0.000016 |            0 |             0 |
| Opening tables       | 0.024121 | 0.004351 |   0.008638 |          800 |             0 |
| init                 | 0.000049 | 0.000021 |   0.000027 |            0 |             0 |
| System lock          | 0.000019 | 0.000008 |   0.000011 |            0 |             0 |
| optimizing           | 0.000022 | 0.000010 |   0.000012 |            0 |             0 |
| statistics           | 0.000030 | 0.000013 |   0.000016 |            0 |             0 |
| preparing            | 0.000026 | 0.000012 |   0.000015 |            0 |             0 |
| executing            | 0.000013 | 0.000005 |   0.000007 |            0 |             0 |
| Sending data         | 3.169882 | 2.755171 |   0.389367 |       534176 |             0 |
| end                  | 0.000069 | 0.000050 |   0.000018 |            0 |             0 |
| query end            | 0.000029 | 0.000022 |   0.000008 |            0 |             0 |
| closing tables       | 0.000031 | 0.000023 |   0.000009 |            0 |             0 |
| freeing items        | 0.000035 | 0.000025 |   0.000009 |            0 |             0 |
| cleaning up          | 0.000038 | 0.000028 |   0.000010 |            0 |             0 |
+----------------------+----------+----------+------------+--------------+---------------+
15 rows in set, 1 warning (0.00 sec)

StoneDB

mysql> set profiling=on;                                                            
Query OK, 0 rows affected, 1 warning (0.00 sec)

mysql> select count(*) from t_user where first_name='柳' and copy_id=9968888;
+----------+
| count(*) |
+----------+
|        1 |
+----------+
1 row in set (0.01 sec)

mysql> show profile cpu,block io for query 1;                                
+----------------------+----------+----------+------------+--------------+---------------+
| Status               | Duration | CPU_user | CPU_system | Block_ops_in | Block_ops_out |
+----------------------+----------+----------+------------+--------------+---------------+
| starting             | 0.000173 | 0.000081 |   0.000086 |            0 |             0 |
| checking permissions | 0.000026 | 0.000013 |   0.000013 |            0 |             0 |
| Opening tables       | 0.010228 | 0.009385 |   0.000843 |            0 |           240 |
| System lock          | 0.000232 | 0.000113 |   0.000119 |            0 |             0 |
| init                 | 0.000045 | 0.000021 |   0.000022 |            0 |             0 |
| optimizing           | 0.000144 | 0.000071 |   0.000074 |            0 |             0 |
| update multi-index   | 0.003694 | 0.002027 |   0.006428 |            0 |             0 |
| aggregation          | 0.000191 | 0.000093 |   0.000098 |            0 |            16 |
| query end            | 0.000020 | 0.000010 |   0.000010 |            0 |             0 |
| closing tables       | 0.000029 | 0.000014 |   0.000015 |            0 |             0 |
| freeing items        | 0.000033 | 0.000016 |   0.000017 |            0 |             0 |
| cleaning up          | 0.000027 | 0.000013 |   0.000013 |            0 |             0 |
+----------------------+----------+----------+------------+--------------+---------------+
12 rows in set, 1 warning (0.00 sec)

从SQL profile可知,SQL在StoneDB执行的过程中,在aggregation阶段发生IO。
综上所述:

  1. 知识网格技术过滤出相关的数据包后,只需要读取元数据,不再解压缩数据包;
  2. 知识网格技术过滤出可疑的数据包后,需要解压缩数据包。

好了,本文到此结束,带大家了解了《【StoneDB技术解析】验证相关数据包是否需要解压缩》,希望本文对你有所帮助!关注golang学习网公众号,给大家分享更多数据库知识!

声明:本文转载于:SegmentFault 如有侵犯,请联系study_golang@163.com删除
相关阅读
更多>
最新阅读
更多>
课程推荐
更多>
评论列表