Question

我有这样的查询（由Hibernate生成 - 为了更清晰而编辑 - 下面你会找到完整的查询）：

select 
visitgroup0_.VISIT_ID as col_0_0_ ... 
from 
ADM_VISIT_GROUP visitgroup0_, 
ADM_CONTACT_OWNER contactown1_ 
where 
visitgroup0_.TIME>'2012-02-14 02:59:24' and 
visitgroup0_.VENDOR_ID='***' and 
visitgroup0_.CONTACT_STATE='PROSPECT' and 
contactown1_.CONTACT_ID=visitgroup0_.CONTACT_ID and 
contactown1_.OWNER_ID='***' order by visitgroup0_.TIME desc limit 30;

简单地说，我正在我们的网站上记录客户访问，每个客户都分配了所有者 - 销售代表。我需要向所有者访问他的联系人。

访问存储在ADM_VISIT_GROUP表中 - ＆gt; 200 000行，访问的contact_id列指向ADM_CONTACT。联系人存储在ADM_CONTACT表中 - 大约500 000行。联系人的所有权存储在ADM_CONTACT_OWNER表中 - 与ADM_CONTACT中的行数或多或少相同。

解释说明：

+----+-------------+--------------+--------+----------------------------------------------------------------------------------------------------------------------------------+----------------------+---------+------------------------------------------------+-------+--------------------------+
| id | select_type | table        | type   | possible_keys                                                                                                                    | key                  | key_len | ref                                            | rows  | Extra                    |
+----+-------------+--------------+--------+----------------------------------------------------------------------------------------------------------------------------------+----------------------+---------+------------------------------------------------+-------+--------------------------+
|  1 | SIMPLE      | visitgroup0_ | range  | FK_VISIT_GROUP_VENDOR_ID,IDX_VISIT_GROUP_VENDOR_ID_CONTACT_ID,IDX_VISIT_GROUP_TIME,IDX_VISIT_GROUP_CONTACT_ID                    | IDX_VISIT_GROUP_TIME | 9       | NULL                                           | 19640 | Using where              |
|  1 | SIMPLE      | contactown1_ | eq_ref | CONTACT_ID,FK_CONTACT_OWNER_CONTACT_ID,FK_CONTACT_OWNER_OWNER_ID,IDX_CONTACT_OWNER_CONTACT_OWNER,IDX_CONTACT_OWNER_OWNER_DELETED | CONTACT_ID           | 1534    | salesmanago_main.visitgroup0_.CONTACT_ID,const |     1 | Using where; Using index |
+----+-------------+--------------+--------+----------------------------------------------------------------------------------------------------------------------------------+----------------------+---------+------------------------------------------------+-------+--------------------------+

使用以下方法创建ADM_VISIT_GROUP表：

CREATE TABLE `ADM_VISIT_GROUP` (
  `ID` char(128) COLLATE utf8_polish_ci NOT NULL,
  `UUID` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `CONTACT_STATE` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `VISIT_ID` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `HOST` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `IP` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `LOCATION` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `URI` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `TIME` datetime DEFAULT NULL,
  `CONVERSATION` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `DURATION` bigint(20) DEFAULT NULL,
  `VISIT_SOURCE` varchar(255) COLLATE utf8_polish_ci NOT NULL,
  `VISIT_SOURCE_HOST` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `VISIT_SOURCE_KEYWORDS` varchar(1024) COLLATE utf8_polish_ci DEFAULT NULL,
  `VISIT_SOURCE_DETAILS` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `VISIT_SCORE` bigint(20) DEFAULT NULL,
  `CLIENT` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `EMAIL` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `CONTACT_ID` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `CONVERSATION_ID` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `IP_ORGANIZATION` varchar(255) COLLATE utf8_polish_ci DEFAULT NULL,
  `ISP_ONLY` tinyint(1) DEFAULT NULL,
  `VENDOR_ID` varchar(255) COLLATE utf8_polish_ci NOT NULL,
  PRIMARY KEY (`ID`),
  KEY `FK_VISIT_GROUP_VENDOR_ID` (`VENDOR_ID`),
  KEY `IDX_VISIT_GROUP_VENDOR_ID_CONTACT_ID` (`VENDOR_ID`,`CONTACT_ID`),
  KEY `IDX_VISIT_GROUP_ISP_ONLY_VENDOR_ID_CONTACT_ID` (`ISP_ONLY`,`VENDOR_ID`,`CONTACT_ID`),
  KEY `IDX_VISIT_GROUP_TIME` (`TIME`),
  KEY `IDX_VISIT_GROUP_EMAIL` (`EMAIL`),
  KEY `IDX_VISIT_GROUP_CONTACT_ID` (`CONTACT_ID`),
  CONSTRAINT `FK_VISIT_GROUP_VENDOR_ID` FOREIGN KEY (`VENDOR_ID`) REFERENCES `ADM_VENDOR` (`ID`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_polish_ci

表ADM_CONTACT_OWNER：

CREATE TABLE `ADM_CONTACT_OWNER` (
  `ID` char(128) COLLATE utf8_polish_ci NOT NULL,
  `VERSION` int(11) NOT NULL,
  `CONTACT_ID` varchar(255) COLLATE utf8_polish_ci NOT NULL,
  `OWNER_ID` varchar(255) COLLATE utf8_polish_ci NOT NULL,
  `CREATED_ON` datetime NOT NULL,
  `OWNERSHIP_RIGHTS` varchar(255) COLLATE utf8_polish_ci NOT NULL,
  `GRANTED_BY_ID` varchar(255) COLLATE utf8_polish_ci NOT NULL,
  `MODIFIED_ON` datetime NOT NULL,
  `DELETED` tinyint(1) NOT NULL,
  PRIMARY KEY (`ID`),
  UNIQUE KEY `CONTACT_ID` (`CONTACT_ID`,`OWNER_ID`),
  KEY `FK_CONTACT_OWNER_GRANTED_BY_ID` (`GRANTED_BY_ID`),
  KEY `FK_CONTACT_OWNER_CONTACT_ID` (`CONTACT_ID`),
  KEY `FK_CONTACT_OWNER_OWNER_ID` (`OWNER_ID`),
  KEY `IDX_CONTACT_OWNER_CONTACT_OWNER` (`CONTACT_ID`,`OWNER_ID`),
  KEY `IDX_CONTACT_OWNER_OWNER_DELETED` (`OWNER_ID`,`DELETED`),
  CONSTRAINT `FK_CONTACT_OWNER_CONTACT_ID` FOREIGN KEY (`CONTACT_ID`) REFERENCES `ADM_CONTACT` (`ID`),
  CONSTRAINT `FK_CONTACT_OWNER_GRANTED_BY_ID` FOREIGN KEY (`GRANTED_BY_ID`) REFERENCES `ADM_USER_ACCOUNT` (`ID`),
  CONSTRAINT `FK_CONTACT_OWNER_OWNER_ID` FOREIGN KEY (`OWNER_ID`) REFERENCES `ADM_USER_ACCOUNT` (`ID`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_polish_ci

完整查询：

    select 
visitgroup0_.VISIT_ID as col_0_0_, 
visitgroup0_.DURATION as col_1_0_, 
visitgroup0_.TIME as col_2_0_, 
visitgroup0_.VISIT_SOURCE as col_3_0_, 
visitgroup0_.VISIT_SOURCE_HOST as col_4_0_, 
visitgroup0_.VISIT_SOURCE_KEYWORDS as col_5_0_, 
visitgroup0_.EMAIL as col_6_0_, 
visitgroup0_.IP_ORGANIZATION as col_7_0_, 
visitgroup0_.CLIENT as col_8_0_, 
visitgroup0_.HOST as col_9_0_, 
visitgroup0_.IP as col_10_0_, 
visitgroup0_.LOCATION as col_11_0_, 
visitgroup0_.URI as col_12_0_, 
visitgroup0_.UUID as col_13_0_, 
visitgroup0_.VISIT_SCORE as col_14_0_, 
visitgroup0_.CONVERSATION as col_15_0_, 
visitgroup0_.CONTACT_ID as col_16_0_, 
visitgroup0_.CONVERSATION_ID as col_17_0_ 
from 
ADM_VISIT_GROUP visitgroup0_, 
ADM_CONTACT_OWNER contactown1_ 
where 
visitgroup0_.TIME>'2012-02-14 02:59:24' and 
visitgroup0_.VENDOR_ID='3739d7a7-2e8f-4409-bd2d-2b505b5e7749' and 
visitgroup0_.CONTACT_STATE='PROSPECT' and 
contactown1_.CONTACT_ID=visitgroup0_.CONTACT_ID and 
contactown1_.OWNER_ID='3f440a3e-a55e-44f5-ac75-d30bd27f4f97' 
order by visitgroup0_.TIME desc limit 30;

任何人都可以帮助我提高查询速度吗？

Answer 1

它必须手动过滤掉很多行，因为mysql只能在TIME上使用索引（IDX_VISIT_GROUP_TIME）

因为MySQL在查询中每个表只能使用一个索引，所以请尝试在上面创建组合索引：

VENDOR_ID
CONTACT_STATE
CONTACT_ID
TIME

列的顺序在索引中很重要！如果所有子句都在查询中，则只能从第一个子句到最后一个子句使用索引。并使用＆gt;或类似的东西很可能会使索引的其余部分无法使用。这就是TIME在索引中排在最后的原因。

但是某个列没有出现在你的where子句中，它无法到达TIME列的索引部分，索引的效果要差很多。这就是为什么在这种情况下你会想要用这些列创建一个单独的索引。

Answer 2

我会：

将OWNER_ID，VENDOR_ID和CONTACT_ID数据类型从VARCHAR更改为UNSIGNED INTEGER或任何其他数字类型
在JOIN和ADM_VISIT_GROUP

ADM_CONTACT_OWNER

将CONTACT_STATE更改为STATE表的外键或使用数字字段
改进INDEX创作

提高查询性能

2 个答案: