我在下面有这个查询，我已经重写了十几种不同的方法，但我无法在8秒内对其进行优化和加载。如果我可以达到2s，那就太好了。 1秒或更少是最佳的。
此查询检索当前可供销售或交易的图书列表，并执行一些过滤。此查询大约需要9-10秒。
SELECT
listing.for_sale,
listing.for_trade,
MIN(listing.price) AS from_price,
MAX(listing.price) AS to_price,
IF (NOW() > CONVERT_TZ(listing.date_sale_expiration, '+00:00', '-7:00'), 1, 0) AS expired,
COUNT(listing.book_id) AS 'listing_count',
book.id AS 'book_id',
book.title AS 'book_title',
book.date_released AS 'date_released',
book.publisher AS 'publisher',
book.photo_cover AS 'photo_cover',
publisher.name AS 'publisher_name',
COALESCE((SELECT COUNT(*) FROM listing l1 WHERE l1.book_id = book.id AND l1.status IN ('in_active_deal', 'complete')), 0) AS 'number_sold',
(SELECT 1 FROM listing l2 WHERE l2.status = 'available' AND l2.book_id = book.id AND l2.member_id = 1234 LIMIT 1) AS 'hasListing',
(SELECT 1 FROM wishlist w1 WHERE w1.book_id = book.id AND w1.member_id = 1234 LIMIT 1) AS 'hasWishlist'
FROM listing
INNER JOIN member ON
listing.member_id = member.id
AND member.transaction_limit <> 0
AND member.banned <> 1
AND member.date_last_login > DATE_SUB(CURDATE(), INTERVAL 120 DAY)
INNER JOIN book ON
listing.book_id = book.id
AND book.released = 1
INNER JOIN publisher ON
book.publisher_id = publisher.id
WHERE
listing.type = 'book'
AND listing.status = 'available'
AND (listing.for_trade = 1 OR (listing.for_sale = 1 AND NOW() < COALESCE(CONVERT_TZ(listing.date_sale_expiration, '+00:00', '-7:00'), 0)))
AND (
EXISTS (SELECT 1 FROM listing l3 LEFT JOIN book b ON l3.book_id = b.id WHERE l3.member_id = 1234 AND b.publisher_id = book.publisher_id AND l3.status = 'available' AND l3.type = 'book' AND (l3.for_trade = 1 OR (l3.for_sale = 1 AND NOW() < COALESCE(CONVERT_TZ(l3.date_sale_expiration, '+00:00', '-7:00'), 0))) LIMIT 1)
OR member.publisher_only <> 1
OR member.id = 1234
)
AND (
EXISTS (SELECT 1 FROM wishlist w WHERE w.member_id = member.id AND w.type = 'book' AND (w.type, w.book_id) IN (SELECT l4.type, l4.book_id FROM listing l4 WHERE 1234 = l4.member_id AND l4.status = 'available' AND (l4.for_trade = 1 OR (l4.for_sale = 1 AND NOW() < COALESCE(DATE_SUB(l4.date_sale_expiration, INTERVAL 7 HOUR), 0)))) LIMIT 1)
OR member.wishlist_only <> 1
OR member.id = 1234
)
GROUP BY
book.id
ORDER BY
book.date_released DESC
LIMIT 30;
这些是我的表格：
CREATE TABLE `listing` (
`id` int(10) unsigned NOT NULL auto_increment,
`member_id` int(10) unsigned NOT NULL,
`type` enum('book','audiobook','accessory') NOT NULL,
`book_id` int(10) unsigned default NULL,
`audiobook_id` int(10) unsigned default NULL,
`accessory_id` int(10) unsigned default NULL,
`date_created` datetime NOT NULL,
`date_modified` datetime NOT NULL,
`date_sale_expiration` datetime default NULL,
`status` enum('available','in_active_deal','complete','deleted') NOT NULL default 'available',
`for_sale` tinyint(1) unsigned NOT NULL default '0',
`for_trade` tinyint(1) unsigned NOT NULL default '0',
`price` decimal(10,2) default NULL,
`condition` tinyint(1) unsigned default NULL,
`notes` varchar(255) default NULL,
PRIMARY KEY (`id`),
KEY `ix_accessory` (`accessory_id`,`member_id`,`type`,`status`),
KEY `ix_book` (`book_id`,`member_id`,`type`,`status`),
KEY `ix_member` (`member_id`,`status`,`date_created`),
KEY `ix_audiobook` (`audiobook_id`,`member_id`,`type`,`status`),
KEY `ix_status` (`status`,`accessory_id`,`for_trade`,`member_id`)
) ENGINE=MyISAM AUTO_INCREMENT=281724 DEFAULT CHARSET=utf8
CREATE TABLE `member` (
`id` int(10) unsigned NOT NULL auto_increment,
`email` varchar(200) NOT NULL,
`screen_name` varchar(25) default NULL,
`date_last_login` datetime default NULL,
`wishlist_only` tinyint(1) unsigned NOT NULL default '1',
`platform_only` tinyint(1) unsigned NOT NULL default '0',
`transaction_limit` smallint(6) NOT NULL default '5',
`banned` tinyint(1) unsigned NOT NULL default '0',
`notes` text,
PRIMARY KEY (`id`),
KEY `ix_email` (`email`),
KEY `ix_screen_name` (`screen_name`),
KEY `ix_transaction_limit` (`transaction_limit`)
) ENGINE=MyISAM AUTO_INCREMENT=50842 DEFAULT CHARSET=utf8
CREATE TABLE `publisher` (
`id` int(10) unsigned NOT NULL auto_increment,
`name` varchar(128) NOT NULL,
`date_updated` datetime default NULL,
PRIMARY KEY (`id`),
KEY `ix_name` (`name`)
) ENGINE=MyISAM AUTO_INCREMENT=129 DEFAULT CHARSET=utf8
CREATE TABLE `book` (
`id` int(10) unsigned NOT NULL auto_increment,
`publisher_id` int(10) unsigned default NULL,
`name` varchar(128) NOT NULL,
`description` text,
`keywords` varchar(200) default NULL,
`date_released` varchar(10) default NULL,
`genre` varchar(50) default NULL,
`subgenre` varchar(50) default NULL,
`author` varchar(100) default NULL,
`date_updated` datetime default NULL,
`photo_cover` varchar(50) default NULL,
`weight_oz` decimal(7,2) default NULL,
`released` tinyint(2) NOT NULL default '0',
PRIMARY KEY (`id`),
KEY `ix_genre` (`genre`),
KEY `ix_name` (`name`),
KEY `ix_released` (`released`,`date_released`),
KEY `ix_keywords` (`keywords`)
) ENGINE=MyISAM AUTO_INCREMENT=87329 DEFAULT CHARSET=utf8
CREATE TABLE `wishlist` (
`id` int(10) unsigned NOT NULL auto_increment,
`member_id` int(10) unsigned NOT NULL,
`type` enum('book','audiobook','accessory') NOT NULL,
`book_id` int(10) unsigned default NULL,
`audiobook_id` int(10) unsigned default NULL,
`accessory_id` int(10) unsigned default NULL,
`date_created` datetime NOT NULL,
`date_modified` datetime NOT NULL,
PRIMARY KEY (`id`),
KEY `ix_accessory` (`accessory_id`,`member_id`,`type`),
KEY `ix_book` (`book_id`,`member_id`,`type`),
KEY `ix_member_accessory` (`member_id`,`accessory_id`),
KEY `ix_member_date_created` (`member_id`,`date_created`),
KEY `ix_member_book` (`member_id`,`book_id`),
KEY `ix_member_audiobook` (`member_id`,`audiobook_id`),
KEY `ix_audiobook` (`audiobook_id`,`member_id`,`type`)
) ENGINE=MyISAM AUTO_INCREMENT=241886 DEFAULT CHARSET=utf8
这是我运行EXPLAIN时的结果：
+----+--------------------+-----------+----------------+---------------------------------------------------------------------------------------+----------------------+---------+------------------------------------+-------+----------------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+--------------------+-----------+----------------+---------------------------------------------------------------------------------------+----------------------+---------+------------------------------------+-------+----------------------------------------------+
| 1 | PRIMARY | member | range | PRIMARY,ix_transaction_limit | ix_transaction_limit | 2 | NULL | 19617 | Using where; Using temporary; Using filesort |
| 1 | PRIMARY | listing | ref | ix_game,ix_member,ix_status | ix_member | 5 | live_database001.member.id,const | 7 | Using where |
| 1 | PRIMARY | book | eq_ref | PRIMARY,ix_released | PRIMARY | 4 | live_database001.listing.book_id | 1 | Using where |
| 1 | PRIMARY | publisher | eq_ref | PRIMARY | PRIMARY | 4 | live_database001.book.publisher_id | 1 | |
| 6 | DEPENDENT SUBQUERY | w | ref | ix_member_accessory,ix_member_date_created,ix_member_book,ix_member_publisher | ix_member_accessory | 4 | live_database001.member.id | 6 | Using where |
| 7 | DEPENDENT SUBQUERY | l4 | index_subquery | ix_book,ix_member,ix_status | ix_book | 11 | func,const,func,const | 1 | Using where |
| 5 | DEPENDENT SUBQUERY | l3 | ref | ix_book,ix_member,ix_status | ix_member | 5 | const,const | 63 | Using where |
| 5 | DEPENDENT SUBQUERY | b | eq_ref | PRIMARY | PRIMARY | 4 | live_database001.l3.book_id | 1 | Using where |
| 4 | DEPENDENT SUBQUERY | w1 | ref | ix_book,ix_member_accessory,ix_member_date_created,ix_member_game,ix_member_publisher | ix_book | 9 | func,const | 1 | Using where; Using index |
| 3 | DEPENDENT SUBQUERY | l2 | ref | ix_book,ix_member,ix_status | ix_book | 9 | func,const | 1 | Using where; Using index |
| 2 | DEPENDENT SUBQUERY | l1 | ref | ix_book,ix_status | ix_book | 5 | func | 10 | Using where; Using index |
+----+--------------------+-----------+----------------+--------------------------------------------------------------------------------------+----------------------+---------+------------------------------------+-------+----------------------------------------------+
这让我有几个问题：
1.成员表正在使用ix_transaction_limit，因此搜索了19k +行。由于我指定的是member.id，因此不应该使用PRIMARY，rows不应该是1吗？我该如何解决这个问题？
2. key_len如何影响表现？
我已经看到其他处理100万行的复杂查询花费的时间更少。怎么只有19k行花了这么长时间？
（我对MySQL优化仍然非常环保，所以我真的很想了解它的原理和原因）
非常感谢任何小或大的建议，提前谢谢！
答案 0 :(得分：0)
不确定什么交易限制，但猜测似乎是一个奇怪的选择索引。可能有帮助的是date_last_login上的索引。目前查询是过滤成员，然后加入列表 - 即。它通过具有适当事务限制的所有成员记录并使用成员ID查找列表。
答案 1 :(得分：0)
在成员表上删除索引后，我仍然遇到同样的问题。事实上，这使情况变得更糟。所以我的最终解决方案是从头开始重新编写查询。
因此，改变条件中的顺序也会产生很大的不同。因此，将and member_id = 1234和and wishlish_only <> 1移到子查询上方是一个巨大的改进。
感谢您的帮助！