我有以下情况，想要选择最佳索引（一些单一索引和一些多索引）。
行号：14,000,000
表：添加
CREATE TABLE `Add` (
`ID` int(11) NOT NULL AUTO_INCREMENT,
`Name` varchar(255) NOT NULL,
`Description` longtext,
`IDUser` int(11) NOT NULL,
`DateAdded` datetime(4) DEFAULT NULL,
`IDCategory` int(11) NOT NULL,
`RowVersion` datetime NOT NULL,
`IDStatus` int(11) DEFAULT '3',
`IDModeration` int(11) DEFAULT '4',
`Price` double DEFAULT NULL,
`DateFrom` datetime DEFAULT NULL,
`DateTo` datetime DEFAULT NULL,
`City` varchar(50) DEFAULT NULL,
`IDRegion` int(11) DEFAULT NULL,
`IDCity` int(11) DEFAULT NULL,
`IDNeighbourhood` int(11) DEFAULT NULL,
`DateAccepted` datetime(4) DEFAULT NULL,
`DateUpdated` datetime DEFAULT NULL,
`latitude` double DEFAULT NULL,
`longitude` double DEFAULT NULL,
`ShowLandLine` tinyint(1) DEFAULT NULL,
`DateUpToDated` datetime(4) DEFAULT NULL,
`Telephone` varchar(50) DEFAULT NULL,
`UserType` tinyint(1) DEFAULT NULL,
`Neighbourhood` varchar(255) DEFAULT NULL,
`IP` varchar(16) DEFAULT NULL,
`FullEditToken` varchar(32) DEFAULT NULL,
`IDSource` tinyint(4) DEFAULT NULL,
`ExpiredAt` datetime DEFAULT NULL,
`BadWordsFlag` tinyint(4) DEFAULT NULL,
PRIMARY KEY (`ID`),
KEY `IX_Add_DateFrom` (`DateFrom`),
KEY `IX_Add_IDUser` (`IDUser`),
KEY `IX_Add_DateUpdated` (`DateUpdated`),
KEY `IX_Add_DateUpToDated` (`DateUpToDated`),
KEY `IX_Add_DateTo` (`DateTo`),
KEY `IX_Add_ON_IDUser_IDStatus_IDModeratiON_DateFrom_DateTo` (`IDUser`,`IDStatus`,`IDModeration`,`DateFrom`,`DateTo`),
KEY `IX_Add_DateAdded` (`DateAdded`),
KEY `IX_Add_Telephone` (`Telephone`),
KEY `FK_Add_AddModeration` (`IDModeration`),
KEY `FK_Add_AddStatus` (`IDStatus`),
KEY `FK_Add_Category` (`IDCategory`),
KEY `FK_Add_col_city` (`IDCity`),
KEY `FK_Add_col_neighbourhood` (`IDNeighbourhood`),
KEY `FK_Add_col_region` (`IDRegion`),
KEY `IDSource` (`IDSource`),
KEY `IX_Add_Name` (`Name`),
KEY `IX_Add_IP` (`IP`),
KEY `IX_Add_UserType` (`UserType`),
KEY `IX_Add_IDUser_IDStatus` (`IDUser`,`IDStatus`),
KEY `FullEditToken` (`FullEditToken`),
KEY `IX_ExpiredAt` (`ExpiredAt`),
CONSTRAINT `Add_ibfk_1` FOREIGN KEY (`IDSource`) REFERENCES `AddSource` (`ID`) ON DELETE NO ACTION ON UPDATE NO ACTION,
CONSTRAINT `FK_Add_AddModeration` FOREIGN KEY (`IDModeration`) REFERENCES `AddModeration` (`ID`) ON DELETE NO ACTION ON UPDATE NO ACTION,
CONSTRAINT `FK_Add_AddStatus` FOREIGN KEY (`IDStatus`) REFERENCES `AddStatus` (`ID`) ON DELETE NO ACTION ON UPDATE NO ACTION,
CONSTRAINT `FK_Add_Category` FOREIGN KEY (`IDCategory`) REFERENCES `Category` (`ID`) ON DELETE NO ACTION ON UPDATE NO ACTION,
CONSTRAINT `FK_Add_User` FOREIGN KEY (`IDUser`) REFERENCES `User` (`ID`) ON DELETE NO ACTION ON UPDATE NO ACTION,
CONSTRAINT `FK_Add_col_city` FOREIGN KEY (`IDCity`) REFERENCES `col_city` (`Id`) ON DELETE NO ACTION ON UPDATE NO ACTION,
CONSTRAINT `FK_Add_col_neighbourhood` FOREIGN KEY (`IDNeighbourhood`) REFERENCES `col_neighbourhood` (`Id`) ON DELETE NO ACTION ON UPDATE NO ACTION,
CONSTRAINT `FK_Add_col_region` FOREIGN KEY (`IDRegion`) REFERENCES `col_region` (`ID`) ON DELETE NO ACTION ON UPDATE NO ACTION
) ENGINE=InnoDB AUTO_INCREMENT=16452993 DEFAULT CHARSET=utf8 |
show index from Add返回：
+-------+------------+--------------------------------------------------------+--------------+-----------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| Table | Non_unique | Key_name | Seq_in_index | Column_name | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment | Index_comment |
+-------+------------+--------------------------------------------------------+--------------+-----------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| Add | 0 | PRIMARY | 1 | ID | A | 1905 | NULL | NULL | | BTREE | | |
| Add | 1 | IX_Add_DateFrom | 1 | DateFrom | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_IDUser | 1 | IDUser | A | 1905 | NULL | NULL | | BTREE | | |
| Add | 1 | IX_Add_DateUpdated | 1 | DateUpdated | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_DateUpToDated | 1 | DateUpToDated | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_DateTo | 1 | DateTo | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_ON_IDUser_IDStatus_IDModeratiON_DateFrom_DateTo | 1 | IDUser | A | 1905 | NULL | NULL | | BTREE | | |
| Add | 1 | IX_Add_ON_IDUser_IDStatus_IDModeratiON_DateFrom_DateTo | 2 | IDStatus | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_ON_IDUser_IDStatus_IDModeratiON_DateFrom_DateTo | 3 | IDModeration | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_ON_IDUser_IDStatus_IDModeratiON_DateFrom_DateTo | 4 | DateFrom | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_ON_IDUser_IDStatus_IDModeratiON_DateFrom_DateTo | 5 | DateTo | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_DateAdded | 1 | DateAdded | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_Telephone | 1 | Telephone | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | FK_Add_AddModeration | 1 | IDModeration | A | 6 | NULL | NULL | YES | BTREE | | |
| Add | 1 | FK_Add_AddStatus | 1 | IDStatus | A | 4 | NULL | NULL | YES | BTREE | | |
| Add | 1 | FK_Add_Category | 1 | IDCategory | A | 190 | NULL | NULL | | BTREE | | |
| Add | 1 | FK_Add_col_city | 1 | IDCity | A | 635 | NULL | NULL | YES | BTREE | | |
| Add | 1 | FK_Add_col_neighbourhood | 1 | IDNeighbourhood | A | 952 | NULL | NULL | YES | BTREE | | |
| Add | 1 | FK_Add_col_region | 1 | IDRegion | A | 63 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IDSource | 1 | IDSource | A | 10 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_Name | 1 | Name | A | 1905 | NULL | NULL | | BTREE | | |
| Add | 1 | IX_Add_IP | 1 | IP | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_UserType | 1 | UserType | A | 4 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_Add_IDUser_IDStatus | 1 | IDUser | A | 1905 | NULL | NULL | | BTREE | | |
| Add | 1 | IX_Add_IDUser_IDStatus | 2 | IDStatus | A | 1905 | NULL | NULL | YES | BTREE | | |
| Add | 1 | FullEditToken | 1 | FullEditToken | A | 127 | NULL | NULL | YES | BTREE | | |
| Add | 1 | IX_ExpiredAt | 1 | ExpiredAt | A | 100 | NULL | NULL | YES | BTREE | | |
+-------+------------+--------------------------------------------------------+--------------+-----------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
27 rows in set (0.00 sec)
我们的主要疑问是
##系统中使用的大多数查询共同部分：
SELECT `a`.`Description` AS `Description`,
GROUP_CONCAT(CONCAT_WS('=', `catt`.`Name`,
IF(`gc`.`Name` IS NOT NULL,
`gc`.`Name`, `aav`.`Value`))) AS `Attributes`,
(( SELECT IF(`ae`.`IDAdmin` IS NOT NULL, CONCAT('Admin=', `adm`.`adminName`),
IF(`ae`.`IDUser` IS NOT NULL, CONCAT('USER=', `ae`.`IDUser`), '') )
FROM `AddEvent` AS `ae`
LEFT JOIN `Admin` AS `adm` ON `ae`.`IDAdmin` = `adm`.`adminId`
WHERE `a`.`ID` = `ae`.`IDAdd`
ORDER BY `ae`.`AddedAt` DESC
LIMIT 1)
) AS `AdminOrUser`,
'accepted' AS `AddStatus`,
`a`.`ID` AS `ID`,
`a`.`Name` AS `Name`, `a`.`IDStatus` AS `IDStatus`, `a`.`BadWordsFlag` AS `BadWordsFlag`,
`a`.`Price` AS `Price`, `u`.`TelephonePrimary` AS `Telephone`,
`a`.`DateTo` AS `DateTo`,
IF(`a`.`DateTo`, `a`.`DateTo`, `a`.`DateAdded`) AS `CustomDate`,
`a`.`IDModeration` AS `IDModeration`,
( ( SELECT 1
FROM `AddChanges` AS `ac`
WHERE `ac`.`IDAdd` = `a`.`ID`
AND `ac`.`IDModeration` = 4
ORDER BY `ac`.`ID` DESC
LIMIT 1) ) AS `HasChange`,
( ( SELECT GROUP_CONCAT(CONCAT_WS('>', `ai`.`Path`, `ai`.`IsDefault`))
FROM `AddImage` AS `ai`
WHERE `ai`.`IDAdd` = `a`.`ID`) ) AS `Images`,
`ca2`.`Name` AS `ParentCategory`,
`ca1`.`Name` AS `Category`,
`r`.`Name` AS `RegionName`,
`ct`.`Name` AS `CityName`,
`n`.`Name` AS `NeighbourhoodName`,
`u`.`Email` AS `Email`
FROM `Add` AS `a`
LEFT JOIN `User` AS `u` ON `u`.`ID` = `a`.`IDUser`
LEFT JOIN `Category` AS `ca1` ON `ca1`.`ID` = `a`.`IDCategory`
LEFT JOIN `Category` AS `ca2` ON `ca2`.`ID` = `ca1`.`IDParent`
LEFT JOIN `col_region` AS `r` ON `r`.`Id` = `a`.`IDRegion`
LEFT JOIN `col_city` AS `ct` ON `ct`.`Id` = `a`.`IDCity`
LEFT JOIN `col_neighbourhood` AS `n` ON `n`.`Id` = `a`.`IDNeighbourhood`
LEFT JOIN `AddAttributeValues` AS `aav` ON `aav`.`IDAdd` = `a`.`ID`
LEFT JOIN `CategoryAttribute` AS `catt` ON `catt`.`ID` = `aav`.`IDCategoryAttribute`
LEFT JOIN `GenericCollection` AS `gc` ON `gc`.`ID` = `aav`.`Value`
＃1详细视图
WHERE (`a`.`DateTo` > CURRENT_TIMESTAMP()
AND `a`.`IDStatus` = 1
AND `a`.`IDModeration` = 1 )
AND (`a`.`IDStatus` <> 2)
GROUP BY `a`.`ID`
ORDER BY `a`.`ID` DESC
LIMIT 51
＃详细状态
WHERE ((`a`.`DateTo` > CURRENT_TIMESTAMP()
AND `a`.`IDStatus` = 1
AND `a`.`IDModeration` = 1
)
AND (`a`.`IDStatus` <> 2)
)
AND (`a`.`IDRegion` = '8')
GROUP BY `a`.`ID`
ORDER BY `a`.`ID` DESC
LIMIT 51
＃详细分类
WHERE ((`a`.`DateTo` > CURRENT_TIMESTAMP()
AND `a`.`IDStatus` = 1
AND `a`.`IDModeration` = 1
)
AND (`a`.`IDStatus` <> 2)
)
AND (`a`.`IDCategory` = '43627'
OR `ca1`.`IDParent` = '43627'
OR `ca1`.`IDParent` IN (
SELECT `ca3`.`ID` AS `ID`
FROM `Category` AS `ca3`
WHERE `ca3`.`IDParent` = '43627')
)
GROUP BY `a`.`ID`
ORDER BY `a`.`ID` DESC
LIMIT 51
#detailed with date category state
WHERE (((((`a`.`DateTo` > CURRENT_TIMESTAMP()
AND `a`.`IDStatus` = 1
AND `a`.`IDModeration` = 1)
AND (`a`.`IDStatus` <> 2)
)
AND (UNIX_TIMESTAMP(`a`.`DateAdded`) >= '1503430200'))
AND (UNIX_TIMESTAMP(`a`.`DateAdded`) <= '1505071799')
)
AND (`a`.`IDCategory` = '43627'
OR `ca1`.`IDParent` = '43627'
OR `ca1`.`IDParent` IN (
SELECT `ca3`.`ID` AS `ID`
FROM `Category` AS `ca3`
WHERE `ca3`.`IDParent` = '43627'))
)
AND (`a`.`IDRegion` = '8')
GROUP BY `a`.`ID`
ORDER BY `a`.`ID` DESC
LIMIT 51
#summer mode
SELECT 'accepted' AS `AddStatus`, `a`.`ID` AS `ID`, `a`.`Name` AS `Name`,
`a`.`IDStatus` AS `IDStatus`, `a`.`BadWordsFlag` AS `BadWordsFlag`,
`a`.`Price` AS `Price`, `u`.`TelephonePrimary` AS `Telephone`,
`a`.`DateTo` AS `DateTo`,
IF(`a`.`DateTo`, `a`.`DateTo`, `a`.`DateAdded`) AS `CustomDate`,
`a`.`IDModeration` AS `IDModeration`,
( ( SELECT 1
FROM `AddChanges` AS `ac`
WHERE `ac`.`IDAdd` = `a`.`ID`
AND `ac`.`IDModeration` = 4
ORDER BY `ac`.`ID` DESC
LIMIT 1)) AS `HasChange`,
( ( SELECT GROUP_CONCAT(CONCAT_WS('>', `ai`.`Path`, `ai`.`IsDefault`))
FROM `AddImage` AS `ai`
WHERE `ai`.`IDAdd` = `a`.`ID`)
) AS `Images`, `ca2`.`Name` AS `ParentCategory`,
`ca1`.`Name` AS `Category`, `r`.`Name` AS `RegionName`,
`ct`.`Name` AS `CityName`, `n`.`Name` AS `NeighbourhoodName`,
`u`.`Email` AS `Email`
FROM `Add` AS `a`
LEFT JOIN `User` AS `u` ON `u`.`ID` = `a`.`IDUser`
LEFT JOIN `Category` AS `ca1` ON `ca1`.`ID` = `a`.`IDCategory`
LEFT JOIN `Category` AS `ca2` ON `ca2`.`ID` = `ca1`.`IDParent`
LEFT JOIN `col_region` AS `r` ON `r`.`Id` = `a`.`IDRegion`
LEFT JOIN `col_city` AS `ct` ON `ct`.`Id` = `a`.`IDCity`
LEFT JOIN `col_neighbourhood` AS `n` ON `n`.`Id` = `a`.`IDNeighbourhood`
WHERE (((((`a`.`DateTo` > CURRENT_TIMESTAMP()
AND `a`.`IDStatus` = 1
AND `a`.`IDModeration` = 1)
AND (`a`.`IDStatus` <> 2))
AND (UNIX_TIMESTAMP(`a`.`DateAdded`) >= '1503430200'))
AND (UNIX_TIMESTAMP(`a`.`DateAdded`) <= '1505071799')
)
AND (`a`.`IDCategory` = '43627'
OR `ca1`.`IDParent` = '43627'
OR `ca1`.`IDParent` IN (
SELECT `ca3`.`ID` AS `ID`
FROM `Category` AS `ca3`
WHERE `ca3`.`IDParent` = '43627'))
)
AND (`a`.`IDRegion` = '8')
ORDER BY `a`.`ID` DESC
LIMIT 51
我不确定索引以及有多少索引以及哪个字段或字段最适合我们的情况，正如我之前提到的，我们在网站上有1400万条广告记录。
更新
我们正在寻找根据我们的查询优化表索引，我们感谢任何线索或解决方案。
查询的 Explain：
*************************** 1. row ***************************
id: 1
select_type: PRIMARY
table: a
type: range
possible_keys: PRIMARY,IX_Add_DateFrom,IX_Add_IDUser,IX_Add_DateUpdated,IX_Add_DateUpToDated,IX_Add_DateTo,IX_Add_ON_IDUser_IDStatus_IDModeratiON_DateFrom_DateTo,IX_Add_DateAdded,IX_Add_Telephone,FK_Add_AddModeration,FK_Add_AddStatus,FK_Add_Category,FK_Add_col_city,FK_Add_col_neighbourhood,FK_Add_col_region,IDSource,IX_Add_Name,IX_Add_IP,IX_Add_UserType,IX_Add_IDUser_IDStatus,FullEditToken,IX_ExpiredAt
key: FK_Add_AddStatus
key_len: 5
ref: NULL
rows: 1480
Extra: Using index condition; Using where
*************************** 2. row ***************************
id: 1
select_type: PRIMARY
table: u
type: eq_ref
possible_keys: PRIMARY
key: PRIMARY
key_len: 4
ref: trumpet.a.IDUser
rows: 1
Extra: NULL
*************************** 3. row ***************************
id: 1
select_type: PRIMARY
table: ca1
type: eq_ref
possible_keys: PRIMARY
key: PRIMARY
key_len: 4
ref: trumpet.a.IDCategory
rows: 1
Extra: NULL
*************************** 4. row ***************************
id: 1
select_type: PRIMARY
table: ca2
type: eq_ref
possible_keys: PRIMARY
key: PRIMARY
key_len: 4
ref: trumpet.ca1.IDParent
rows: 1
Extra: NULL
*************************** 5. row ***************************
id: 1
select_type: PRIMARY
table: r
type: eq_ref
possible_keys: PRIMARY
key: PRIMARY
key_len: 4
ref: trumpet.a.IDRegion
rows: 1
Extra: NULL
*************************** 6. row ***************************
id: 1
select_type: PRIMARY
table: ct
type: eq_ref
possible_keys: PRIMARY
key: PRIMARY
key_len: 4
ref: trumpet.a.IDCity
rows: 1
Extra: NULL
*************************** 7. row ***************************
id: 1
select_type: PRIMARY
table: n
type: eq_ref
possible_keys: PRIMARY
key: PRIMARY
key_len: 4
ref: trumpet.a.IDNeighbourhood
rows: 1
Extra: NULL
*************************** 8. row ***************************
id: 1
select_type: PRIMARY
table: aav
type: ref
possible_keys: IX_AddAttributeValues_IDAdd
key: IX_AddAttributeValues_IDAdd
key_len: 4
ref: trumpet.a.ID
rows: 1
Extra: NULL
*************************** 9. row ***************************
id: 1
select_type: PRIMARY
table: catt
type: eq_ref
possible_keys: PRIMARY
key: PRIMARY
key_len: 4
ref: trumpet.aav.IDCategoryAttribute
rows: 1
Extra: NULL
*************************** 10. row ***************************
id: 1
select_type: PRIMARY
table: gc
type: eq_ref
possible_keys: PRIMARY,IX_GenericCollectiON_ON_IDType_INCLUDE_ID_Name_IDCategory_SortOr
key: PRIMARY
key_len: 4
ref: trumpet.aav.Value
rows: 1
Extra: Using where
*************************** 11. row ***************************
id: 4
select_type: DEPENDENT SUBQUERY
table: ai
type: ref
possible_keys: IX_AddImage_IDAdd
key: IX_AddImage_IDAdd
key_len: 4
ref: func
rows: 1
Extra: NULL
*************************** 12. row ***************************
id: 3
select_type: DEPENDENT SUBQUERY
table: ac
type: ref
possible_keys: IDAdd,IDModeration
key: IDAdd
key_len: 4
ref: func
rows: 1
Extra: Using where; Using filesort
*************************** 13. row ***************************
id: 2
select_type: DEPENDENT SUBQUERY
table: ae
type: ref
possible_keys: IDAdd
key: IDAdd
key_len: 4
ref: func
rows: 3
Extra: Using where; Using filesort
*************************** 14. row ***************************
id: 2
select_type: DEPENDENT SUBQUERY
table: adm
type: eq_ref
possible_keys: PRIMARY
key: PRIMARY
key_len: 4
ref: trumpet.ae.IDAdmin
rows: 1
Extra: NULL
14 rows in set (0.00 sec)
答案 0 :(得分：3)
这不是精确查询只是一个伪代码
然后我们不一定能给你最好的指数。例如，LONGTEXT是SELECT *的一部分会有很大的不同。这是因为该值可能一个额外的磁盘被击中。
左连接到位置，左连接到用户
我很确定逗号是语法错误。如果没有ON子句，我怀疑查询不完整。
请使示例代码中的所有列符合条件。例如，在查询＃1中，无论status和PostDate是否存在巨大差异在同一张桌子里。
您提供的一个表格中没有status或PostDate;请提供其他表格。
status = 2可能是可以优化的，但status <> 2更复杂。这是什么？
我看到了
... AND `a`.`IDStatus` = 1
... AND (`a`.`IDStatus` <> 2) -- Get rid of this?
请解决这些问题。
与此同时，这里有一些试探性的答案
INDEX(status, PostDate) -- in that order (#1)
INDEX(IDModeration, IDStatus, DateTo) -- DateTo last (in big query)
Find the 51 IDs first, _then_ `LEFT JOIN` to the other stuff.
改变这个：
SELECT 1
FROM `AddChanges` AS `ac`
WHERE `ac`.`IDAdd` = `a`.`ID`
AND `ac`.`IDModeration` = 4
ORDER BY `ac`.`ID` DESC
LIMIT 1
到
EXISTS ( SELECT 1 FROM AddChanges
WHERE IdAdd = a.ID
AND IDModeration = 4 )
一个混乱的问题，可以帮助一些人：使用中的许多数据类型都是不必要的大。例如，您真的需要多达20亿个区域吗？缩小数据集会缩小数据集，这会使其更易于缓存，因此I / O更少，因此更快。
低基数列（IDModeration，IDStatus，UserType）上的单列索引往往不会被使用。
过度规范化可能代价高昂（IDModeration，IDStatus）。也就是说，拥有一个状态表而不是字符串的ENUM有什么好处？此外，ENUM为1个字节，而INT为4个字节。
<强>尺寸
使用64GB RAM和14GB数据时，innodb_buffer_pool_size应设置为14GB以上，但不超过50GB;当然不是一个小128MB。并在那时设置innodb_buffer_pool_size = 16。这些变化将提供一些性能，也许很多。 （很难具体说明改进程度。）
特定查询
＃1详细视图。将IDStatus上的当前单列索引替换为
INDEX(IDStatus, IDModeration, DateTo, ID)
详情请见州。将IDRegion上的单列索引替换为：
INDEX(IDRegion, IDStatus, IDModeration, DateTo, ID)
有关类别的详细信息，第一个索引会有所帮助。但也要改变
OR `ca1`.`IDParent` IN (
SELECT `ca3`.`ID` AS `ID`
FROM `Category` AS `ca3`
WHERE `ca3`.`IDParent` = '43627')
到
OR EXISTS ( SELECT 1 FROM `Category` AS `ca3`
WHERE ca1.ID = ID
AND ca1.IDParent = 43627 )
和Category需要INDEX(IDParent, ID)，除非它目前有PRIMARY KEY(ID)。
有关日期类别状态的详细信息，请勿“隐藏函数中的列”。特别是，将UNIX_TIMESTAMP(a.DateAdded) >= '1503430200'更改为
a.DateAdded >= FROM_UNIXTIME(1503430200)
或（等效）
a.DateAdded >= '2017-08-22 12:30:00'
对于摘要模式，上述修复可能会有所帮助。
这一切都可以做到吗？我对此表示怀疑。但是，在深入挖掘之前，我希望您能够实现上述变化。然后我们可以逐步处理下一步看似最好的事情。
提供EXPLAIN时，请使用;，而不是\G;我宁愿挣扎“太宽”而不是“太高”。
您即将达到问题中文本数量的StackOverflow限制。
答案 1 :(得分：1)
create index idx_add
on Ads(LocationID,status,PostDate,Id);
我认为如果使用主键连接所有其他表，则此索引就足够了，您还可以在Location.state＆amp;上创建创建索引。 Category.name
答案 2 :(得分：0)
这些查询已经过优化。您可以创建带索引的临时表并加入它们，而不是创建子查询。 我怀疑使用索引会给你带来显着的性能变化。
如果我非常渴望优化，我在这里唯一要做的就是实现查询。看起来更改的参数是区域，类别，日期范围，我怀疑你真的关心日期范围的几分钟滞后。
这样可以实现您的查询并以最大sql速度获取？