我对我的系统进行了非规范化，并从一个需要数小时的查询转到需要4秒的查询。该表有338,293行。
如果我删除组，总计需要.4秒;它快得多;但后来我必须在Web服务器上处理数千行，并且必须发回大量数据。
是否还有我可以做的索引或更快的方法而不使用文件排序或临时表？
下面是查询和架构。
EXPLAIN
SELECT item_name, category_id, category_name as category, sum(phppos_sales_items.subtotal) as subtotal, sum(phppos_sales_items.total) as total, sum(phppos_sales_items.tax) as tax, sum(phppos_sales_items.profit) as profit, sum(phppos_sales_items.quantity_purchased) as item_sold
FROM `phppos_sales_items` FORCE INDEX (sales_search)
WHERE `phppos_sales_items`.`sale_time` BETWEEN '2016-01-01 00:00:00' and '2016-12-31 23:59:59' and
`phppos_sales_items`.`location_id` IN (1) and
`phppos_sales_items`.`suspended` != 2
AND `deleted` =0
GROUP BY `category_id`
LIMIT 500
+----+-------------+--------------------+-------+----------------------------------------+--------------+---------+------+--------+-------------------------------------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+--------------------+-------+----------------------------------------+--------------+---------+------+--------+-------------------------------------------------------------------+
| 1 | SIMPLE | phppos_sales_items | range | phppos_sales_items_ibfk_5,sales_search | sales_search | 13 | NULL | 169146 | Using index condition; Using MRR; Using temporary; Using filesort |
+----+-------------+--------------------+-------+----------------------------------------+--------------+---------+------+--------+-------------------------------------------------------------------+
1 row in set (0.00 sec)
和...
mysql> show create table phppos_sales_items;
CREATE TABLE `phppos_sales_items` (
`sale_id` int(10) NOT NULL DEFAULT '0',
`item_id` int(10) NOT NULL DEFAULT '0',
`rule_id` int(10) DEFAULT NULL,
`rule_discount` decimal(23,10) DEFAULT NULL,
`description` varchar(255) COLLATE utf8_unicode_ci DEFAULT NULL,
`serialnumber` varchar(255) COLLATE utf8_unicode_ci DEFAULT NULL,
`line` int(11) NOT NULL DEFAULT '0',
`quantity_purchased` decimal(23,10) NOT NULL DEFAULT '0.0000000000',
`item_cost_price` decimal(23,10) NOT NULL,
`item_unit_price` decimal(23,10) NOT NULL,
`regular_item_unit_price_at_time_of_sale` decimal(23,10) DEFAULT NULL,
`discount_percent` decimal(15,3) NOT NULL DEFAULT '0.000',
`commission` decimal(23,10) NOT NULL DEFAULT '0.0000000000',
`is_bogo` tinyint(1) NOT NULL DEFAULT '0',
`sale_time` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP,
`suspended` int(1) NOT NULL DEFAULT '0',
`was_layaway` int(1) NOT NULL DEFAULT '0',
`was_estimate` int(1) NOT NULL DEFAULT '0',
`store_account_payment` int(1) NOT NULL DEFAULT '0',
`location_id` int(11) DEFAULT NULL,
`deleted` int(1) NOT NULL DEFAULT '0',
`item_name` varchar(255) COLLATE utf8_unicode_ci NOT NULL,
`category_name` varchar(255) COLLATE utf8_unicode_ci NOT NULL,
`category_id` int(11) DEFAULT NULL,
`subtotal` decimal(23,10) NOT NULL DEFAULT '0.0000000000',
`tax` decimal(23,10) NOT NULL DEFAULT '0.0000000000',
`total` decimal(23,10) NOT NULL DEFAULT '0.0000000000',
`profit` decimal(23,10) NOT NULL DEFAULT '0.0000000000',
PRIMARY KEY (`sale_id`,`item_id`,`line`),
KEY `item_id` (`item_id`),
KEY `phppos_sales_items_ibfk_3` (`rule_id`),
KEY `item_name` (`item_name`),
KEY `phppos_sales_items_ibfk_5` (`category_id`),
KEY `sales_search` (`sale_time`,`location_id`,`deleted`,
`suspended`,`was_layaway`,`was_estimate`,
`store_account_payment`,`category_id`),
CONSTRAINT `phppos_sales_items_ibfk_1` FOREIGN KEY (`item_id`) REFERENCES `phppos_items` (`item_id`),
CONSTRAINT `phppos_sales_items_ibfk_2` FOREIGN KEY (`sale_id`) REFERENCES `phppos_sales` (`sale_id`),
CONSTRAINT `phppos_sales_items_ibfk_3` FOREIGN KEY (`rule_id`) REFERENCES `phppos_price_rules` (`id`),
CONSTRAINT `phppos_sales_items_ibfk_4` FOREIGN KEY (`location_id`) REFERENCES `phppos_sales` (`location_id`),
CONSTRAINT `phppos_sales_items_ibfk_5` FOREIGN KEY (`category_id`) REFERENCES `phppos_categories` (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci |
1 row in set (0.00 sec)
索引状态：
mysql> mysql> show index from phppos_sales_items WHERE Key_name='sales_search';
+--------------------+------------+--------------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| Table | Non_unique | Key_name | Seq_in_index | Column_name | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment | Index_comment |
+--------------------+------------+--------------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| phppos_sales_items | 1 | sales_search | 1 | deleted | A | 2 | NULL | NULL | | BTREE | | |
| phppos_sales_items | 1 | sales_search | 2 | location_id | A | 2 | NULL | NULL | YES | BTREE | | |
| phppos_sales_items | 1 | sales_search | 3 | suspended | A | 6 | NULL | NULL | | BTREE | | |
| phppos_sales_items | 1 | sales_search | 4 | sale_time | A | 341905 | NULL | NULL | | BTREE | | |
| phppos_sales_items | 1 | sales_search | 5 | category_id | A | 341905 | NULL | NULL | YES | BTREE | | |
+--------------------+------------+--------------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
答案 0 :(得分：1)
所以我的回答是。我用一个带有1.000.000行的样本表进行了测试，它使用前6秒和更改后125毫秒
1）从查询中删除FORCE INDEX
2）添加更好的密钥
ALTER TABLE phppos_sales_items
add KEY `sales_search2` (`deleted`,`location_id`,`suspended`,`sale_time`,`category_id`);
答案 1 :(得分：0)
缩小尺寸会有所帮助，也许很多。你真的有用utf8编码的255个字符的序列号吗？您有佣金高达9,999,999,999,999.9999999999（或欧元，或其他）？等等INT总是4个字节，即使你说INT(1);见TINYINT，即1个字节。
不要使用“范围”启动索引，将该列放在末尾。对于您引用的查询，这会更好：
INDEX(deleted, location_id, sale_time)
你关心你看哪些类别？如果没有ORDER BY，则无法保证您将获得500个。
您遇到了文件或大范围扫描;目前还不清楚哪个更糟糕。您可以添加此选项以为优化程序提供选择：
INDEX(deleted, location_id, category_id)
是的，您可以扩展我建议“覆盖”的任何一个索引;这可能会有所帮助。但重要的是从我提到的列开始，而不是sale_time。
分析分析
sale_id可以是MEDIUMINT UNSIGNED（3个字节）
item_id可以是SMALLINT UNSIGNED（2个字节）
等
不要将ENUM用于明显的数值
不要使用CHAR;坚持VARCHAR
考虑删除always-NULL列
你真的需要10位小数吗？
-88004009373.3500000000的利润？什么？？