Home > Database > Mysql Tutorial > MySQL优化 之 Discuz论坛优化

MySQL优化 之 Discuz论坛优化

WBOY
Release: 2016-06-07 15:26:02
Original
1221 people have browsed it

作/译者:叶金荣(Email: ),来源:http://imysql.cn 一. 前言 近日由于需要,对discuz论坛(简称dz)进行优化,当然了,只是涉及到数据库的优化. 先说一下服务器及dz的数据量,2 * Intel(R) Xeon(TM) CPU 2.40GHz, 4GB mem, SCISC硬盘. MySQL 版本为 4.0.23. 数

作/译者:叶金荣(Email: MySQL优化 之 Discuz论坛优化),来源:http://imysql.cn

一. 前言
近日由于需要,对discuz论坛(简称dz)进行优化,当然了,只是涉及到数据库的优化.
先说一下服务器及dz的数据量,2 * Intel(R) Xeon(TM) CPU 2.40GHz, 4GB mem, SCISC硬盘.
MySQL 版本为 4.0.23. 数据表情况:
cdb_attachments 2万
cdb_members 10万
cdb_posts 68万
cdb_threads 7万

二. 缓存优化
在 my.cnf 中添加/修改以下选项:

 #取消文件系统的外部锁<br> skip-locking<br><br> #不进行域名反解析,注意由此带来的权限/授权问题<br> skip-name-resolve<br><br> #索引缓存,根据内存大小而定,如果是独立的db服务器,可以设置高达80%的内存总量<br> key_buffer = 512M<br><br> #连接排队列表总数<br> back_log = 200<br> max_allowed_packet = 2M<br><br> #打开表缓存总数,可以避免频繁的打开数据表产生的开销<br> table_cache = 512<br><br> #每个线程排序所需的缓冲<br> sort_buffer_size = 4M<br><br> #每个线程读取索引所需的缓冲<br> read_buffer_size = 4M<br><br> #MyISAM表发生变化时重新排序所需的缓冲<br> myisam_sort_buffer_size = 64M<br><br> #缓存可重用的线程数<br> thread_cache = 128<br><br> #查询结果缓存<br> query_cache_size = 128M<br><br> #设置超时时间,能避免长连接<br> set-variable = wait_timeout=60<br><br> #最大并发线程数,cpu数量*2<br> thread_concurrency = 4<br><br> #记录慢查询,然后对慢查询一一优化<br> log-slow-queries = slow.log<br> long_query_time = 1<br><br> #关闭不需要的表类型,如果你需要,就不要加上这个<br> skip-innodb<br> skip-bdb<br> 
Copy after login

以上参数根据各自服务器的配置差异进行调整,仅作为参考.

三. 索引优化
上面提到了,已经开启了慢查询,那么接下来就要对慢查询进行逐个优化了.

1. 搜索优化
搜索的查询SQL大致如下:

 SELECT t.* FROM cdb_posts p, cdb_threads t WHERE<br>    t.fid IN ('37', '45', '4', '6', '17', '41', '28', '32', '31', '1', '42')<br>    AND p.tid=t.tid AND p.author LIKE 'JoansWin'<br>    GROUP BY t.tid ORDER BY lastpost DESC LIMIT 0, 80;<br> 
Copy after login

用 EXPLAIN 分析的结果如下:

 mysql>EXPLAIN  SELECT t.* FROM cdb_posts p, cdb_threads t WHERE<br>    t.fid IN ('37', '45', '4', '6', '17', '41', '28', '32', '31', '1', '42')<br>    AND p.tid=t.tid AND p.author LIKE 'JoansWin'<br>    GROUP BY t.tid ORDER BY lastpost DESC LIMIT 0, 80; <br>+-----------+------------+----------+--------------+-------------+-----------+-------------+<br>| id | select_type | table | type  | possible_keys | key  | key_len | ref         | rows  | Extra<br>+-----------+------------+----------+--------------+-------------+-----------+-------------+<br>|  1 | SIMPLE      | t     | range | PRIMARY,fid   | fid  | 2       | NULL        | 66160 | Using where; <br>Using temporary; Using filesort |<br>|  1 | SIMPLE      | p     | ref   | tid           | tid  | 3       | Forum.t.tid   |    10 | Using where<br>| +----+-------------+-------+-------+---------------+------+---------+-------------+-------+<br>---------<br> 
Copy after login

只用到了 t.fidp.tid,而 p.author 则没有索引可用,总共需要扫描
66160*10 = 661600 次索引,够夸张吧 :(

再分析 cdb_threadscdb_posts 的索引情况:

 mysql>show index from cdb_posts; <br>+-----------+------------+----------+--------------+-------------+-----------+----------<br>---+----------+--------+------+--+<br>| Table     | Non_unique | Key_name | Seq_in_index | Column_name | Collation | Cardinality | Sub_part | <br>Packed | Null | Index_type | Comment | +-----------+------------+----------+--------------+----<br>---------+-----------+-------------+----------+--------+------+--+<br>| cdb_posts |          0 | PRIMARY  |            1 | pid         | A         |      680114 |     NULL | NULL   |<br>| BTREE      |         |<br>| cdb_posts |          1 | fid      |            1 | fid         | A         |          10 |     NULL | NULL   |<br>| BTREE      |         |<br>| cdb_posts |          1 | tid      |            1 | tid         | A         |       68011 |     NULL | NULL   |<br>| BTREE      |         |<br>| cdb_posts |          1 | tid      |            2 | dateline    | A         |      680114 |     NULL | NULL   |<br>| BTREE      |         |<br>| cdb_posts |          1 | dateline |            1 | dateline    | A         |      680114 |     NULL | NULL   |<br>| BTREE      |         | <br>+-----------+------------+----------+--------------+-------------+-----------+---<br> 
Copy after login

以及

 mysql>show index from cdb_threads; <br>+-----------+------------+----------+--------------+-------------+-----------+-------------+<br>----------+--------+------+-----+<br>| Table       | Non_unique | Key_name  | Seq_in_index | Column_name | Collation | Cardinality | Sub_part |<br> Packed | Null | Index_type | Comment | +-----------+------------+----------+--------------+-----<br>--------+-----------+-------------+----------+--------+------+-----+<br>| cdb_threads |          0 | PRIMARY   |            1 | tid         | A         |       68480 |     NULL | NULL   |<br>| BTREE      |         |<br>| cdb_threads |          1 | lastpost  |            1 | topped      | A         |           4 |     NULL | NULL   |<br>| BTREE      |         |<br>| cdb_threads |          1 | lastpost  |            2 | lastpost    | A         |       68480 |     NULL | NULL   |<br>| BTREE      |         |<br>| cdb_threads |          1 | lastpost  |            3 | fid         | A         |       68480 |     NULL | NULL   |<br>| BTREE      |         |<br>| cdb_threads |          1 | replies   |            1 | replies     | A         |         233 |     NULL | NULL   |<br>| BTREE      |         |<br>| cdb_threads |          1 | dateline  |            1 | dateline    | A         |       68480 |     NULL | NULL   |<br>| BTREE      |         |<br>| cdb_threads |          1 | fid       |            1 | fid         | A         |          10 |     NULL | NULL   |<br>| BTREE      |         |<br>| cdb_threads |          1 | enablehot |            1 | enablehot   | A         |           2 |     NULL | NULL   |<br>| BTREE      |         | +-------------+------------+-----------+--------------+-------------+------<br> 
Copy after login

看到索引 fidenablehot 基数太小,看来该索引完全没必要,不过,对于fid基数较大的情况,则可能需要保留>该索引.
所做修改如下:

 ALTER TABLE `cdb_threads` DROP INDEX `enablehot`, DROP INDEX `fid`, ADD INDEX (`fid`, `lastpost`);<br> ALTER TABLE `cdb_posts` DROP INDEX `fid`, ADD INDEX (`author`(10));<br> OPTIMIZE TABLE `cdb_posts`;<br> OPTIMIZE TABLE `cdb_threads`;<br> 
Copy after login

在这里, p.author 字段我设定的部分索引长度是 10, 是我经过分析后得出来的结果,不同的系统,这里的长度也不同,最好自己先取一下平均值,然后再适当调整.
现在,再来执行一次上面的慢查询,发现时间已经从 6s 变成 0.19s,提高了 30 倍.
这次先到这里,下次继续 ^_^

Related labels:
source:php.cn
Statement of this Website
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn
Popular Tutorials
More>
Latest Downloads
More>
Web Effects
Website Source Code
Website Materials
Front End Template