MYSQL OR vs IN性能
我想知道下面的performance是否有差异
SELECT ... FROM ... WHERE someFIELD IN(1,2,3,4) SELECT ... FROM ... WHERE someFIELD between 0 AND 5 SELECT ... FROM ... WHERE someFIELD = 1 OR someFIELD = 2 OR someFIELD = 3 ... 还是将MySQL与编译器优化代码的方式相同？
编辑：由于评论中所述的原因，将AND更改为OR 。
我需要知道这一点，所以我对这两种方法进行了基准testing。 我一致认为IN要比使用OR快得多。
不要相信给自己“意见”的人，科学就是考验和证据。
我运行了一个循环的1000倍的等价查询（为了一致性，我用sql_no_cache ）：
IN ：2.34969592094s
OR ：5.83781504631s
更新：
（我没有原始testing的源代码，就像6年前一样，虽然它返回的结果与本次testing的testing结果相同）
在请求一些示例代码来testing这个，这里是最简单的用例。 使用Eloquent来简化语法，原始的SQL等价物执行相同的操作。
$t = microtime(true); for($i=0; $i<10000; $i++): $q = DB::table('users')->where('id',1) ->orWhere('id',2) ->orWhere('id',3) ->orWhere('id',4) ->orWhere('id',5) ->orWhere('id',6) ->orWhere('id',7) ->orWhere('id',8) ->orWhere('id',9) ->orWhere('id',10) ->orWhere('id',11) ->orWhere('id',12) ->orWhere('id',13) ->orWhere('id',14) ->orWhere('id',15) ->orWhere('id',16) ->orWhere('id',17) ->orWhere('id',18) ->orWhere('id',19) ->orWhere('id',20)->get(); endfor; $t2 = microtime(true); echo $t."\n".$t2."\n".($t2-$t)."\n";
1482080514.3635
1482080517.3713
3.0078368186951
$t = microtime(true); for($i=0; $i<10000; $i++): $q = DB::table('users')->whereIn('id',[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20])->get(); endfor; $t2 = microtime(true); echo $t."\n".$t2."\n".($t2-$t)."\n";
1482080534.0185
1482080536.178
2.1595389842987
我也为未来的Google员工做了testing。 返回的结果总数是10000个中的7264个
SELECT * FROM item WHERE id = 1 OR id = 2 ... id = 10000
这个查询花了0.1239秒
SELECT * FROM item WHERE id IN (1,2,3,...10000)
这个查询花了0.0433秒
IN比OR快3倍
我认为BETWEEN会更快，因为它应该被转换成：
Field >= 0 AND Field <= 5
我的理解是，一个IN将被转换成一堆OR语句。 IN的价值在于易用性。 （保存必须多次input每个列的名称，也使现有的逻辑更容易使用 – 您不必担心AND / OR优先级，因为IN是一个语句。有一堆的OR语句，以确保你括住括号，以确保他们被评估为一个条件。）
你的问题唯一真正的答案是你的问题。 那么你会知道什么在你的特定情况下最好的。
这取决于你在做什么; 范围有多宽，数据types是什么（我知道你的例子使用数字数据types，但是你的问题也可以应用于很多不同的数据types）。
这是一个你想用两种方式写查询的例子; 得到它的工作，然后使用EXPLAIN来找出执行的差异。
我确信这个问题有一个具体的答案，但实际上我是这样想出来的。
这可能有一些帮助： http : //forge.mysql.com/wiki/Top10SQLPerformanceTips
问候，
坦率
我认为sunseeker观察的一个解释是MySQL实际上在IN语句中sorting值，如果它们全部是静态值并且使用二进制search，这比普通的OR方法更有效。 我不记得自己在哪里读过，但sunseeker的结果似乎是一个certificate。
OR会最慢。 无论IN还是BETWEEN更快取决于你的数据，但我期望BETWEEN正常更快，因为它可以简单地从一个索引（假设someField被索引）范围。
我敢打赌，他们是一样的，你可以通过执行以下操作来进行testing：
循环“in（1,2,3,4）”500次，看看需要多长时间。 在“= 1或= 2或= 3 …”版本上循环500次，看看它运行多长时间。
你也可以尝试一个连接的方式，如果someField是一个索引，你的表是大的可能会更快…
SELECT ... FROM ... INNER JOIN (SELECT 1 as newField UNION ALL SELECT 2 UNION ALL SELECT 3 UNION ALL SELECT 4) dt ON someFIELD =newField
我尝试了上面的SQL Server上的join方法，它和in（1,2,3,4）几乎一样，并且都导致了聚集索引seek。 我不确定MySQL将如何处理它们。
从我所理解的编译器优化这些types的查询的方式来看，使用IN子句比多个OR子句更有效率。 如果您有可以使用BETWEEN子句的值，那么效率更高。
我知道，只要你有一个Field的索引，BETWEEN就会用它来快速find一端，然后遍历另一端。 这是最有效的。
我所看到的每一个解释都显示“IN（…）”和“… OR …”可以互换，并且效率相同。 你会期望，因为优化器没有办法知道他们是否包含一个时间间隔。 这也相当于UNION ALL SELECT对各个值的影响。
以下是使用MySQL 5.6 @SQLFiddle的6个查询的详细信息
总之，这6个查询分别覆盖索引列，每个数据types使用2个查询。 所有查询导致使用索引，而不pipe使用IN（）或OR。
| ORs | IN() integer | uses index | uses index date | uses index | uses index varchar | uses index | uses index
我真的只是想揭穿这个OR意味着没有索引可以使用。 这是不正确的。 在以下示例显示中，索引可用于使用OR的查询作为6个查询。
另外在我看来，许多人已经忽略了IN（）是一组OR的语法快捷方式的事实。 在小规模性能之间使用IN（）-v-或者极其（无限）边际。
在更大的范围内，IN（）肯定更方便，但它在逻辑上等同于一组OR条件。 每个查询的环境变化，所以在你的表上testing你的查询总是最好的。
6个解释计划摘要，全部“使用索引条件”（向右滚动）
Query select_type table type possible_keys key key_len ref rows filtered Extra ------------- --------- ------- --------------- ----------- --------- ----- ------ ---------- ----------------------- Integers using OR SIMPLE mytable range aNum_idx aNum_idx 4 10 100.00 Using index condition Integers using IN SIMPLE mytable range aNum_idx aNum_idx 4 10 100.00 Using index condition Dates using OR SIMPLE mytable range aDate_idx aDate_idx 6 7 100.00 Using index condition Dates using IN SIMPLE mytable range aDate_idx aDate_idx 6 7 100.00 Using index condition Varchar using OR SIMPLE mytable range aName_idx aName_idx 768 10 100.00 Using index condition Varchar using IN SIMPLE mytable range aName_idx aName_idx 768 10 100.00 Using index condition
SQL小提琴
MySQL 5.6架构设置 ：
CREATE TABLE `myTable` ( `id` mediumint(8) unsigned NOT NULL auto_increment, `aName` varchar(255) default NULL, `aDate` datetime, `aNum` mediumint(8), PRIMARY KEY (`id`) ) AUTO_INCREMENT=1; ALTER TABLE `myTable` ADD INDEX `aName_idx` (`aName`); ALTER TABLE `myTable` ADD INDEX `aDate_idx` (`aDate`); ALTER TABLE `myTable` ADD INDEX `aNum_idx` (`aNum`); INSERT INTO `myTable` (`aName`,`aDate`) VALUES ("Daniel","2017-09-19 01:22:31") ,("Quentin","2017-06-03 01:06:45") ,("Chester","2017-06-14 17:49:36") ,("Lev","2017-08-30 06:27:59") ,("Garrett","2018-10-04 02:40:37") ,("Lane","2017-01-22 17:11:21") ,("Chaim","2017-09-20 11:13:46") ,("Kieran","2018-03-10 18:37:26") ,("Cedric","2017-05-20 16:25:10") ,("Conan","2018-07-10 06:29:39") ,("Rudyard","2017-07-14 00:04:00") ,("Chadwick","2018-08-18 08:54:08") ,("Darius","2018-10-02 06:55:56") ,("Joseph","2017-06-19 13:20:33") ,("Wayne","2017-04-02 23:20:25") ,("Hall","2017-10-13 00:17:24") ,("Craig","2016-12-04 08:15:22") ,("Keane","2018-03-12 04:21:46") ,("Russell","2017-07-14 17:21:58") ,("Seth","2018-07-25 05:51:30") ,("Cole","2018-06-09 15:32:53") ,("Donovan","2017-08-12 05:21:35") ,("Damon","2017-06-27 03:44:19") ,("Brian","2017-02-01 23:35:20") ,("Harper","2017-08-25 04:29:27") ,("Chandler","2017-09-30 23:54:06") ,("Edward","2018-07-30 12:18:07") ,("Curran","2018-05-23 09:31:53") ,("Uriel","2017-05-08 03:31:43") ,("Honorato","2018-04-07 14:57:53") ,("Griffin","2017-01-07 23:35:31") ,("Hasad","2017-05-15 05:32:41") ,("Burke","2017-07-04 01:11:19") ,("Hyatt","2017-03-14 17:12:28") ,("Brenden","2017-10-17 05:16:14") ,("Ryan","2018-10-10 08:07:55") ,("Giacomo","2018-10-06 14:21:21") ,("James","2018-02-06 02:45:59") ,("Colt","2017-10-10 08:11:26") ,("Kermit","2017-09-18 16:57:16") ,("Drake","2018-05-20 22:08:36") ,("Berk","2017-04-16 17:39:32") ,("Alan","2018-09-01 05:33:05") ,("Deacon","2017-04-20 07:03:05") ,("Omar","2018-03-02 15:04:32") ,("Thaddeus","2017-09-19 04:07:54") ,("Troy","2016-12-13 04:24:08") ,("Rogan","2017-11-02 00:03:25") ,("Grant","2017-08-21 01:45:16") ,("Walker","2016-11-26 15:54:52") ,("Clarke","2017-07-20 02:26:56") ,("Clayton","2018-08-16 05:09:29") ,("Denton","2018-08-11 05:26:05") ,("Nicholas","2018-07-19 09:29:55") ,("Hashim","2018-08-10 20:38:06") ,("Todd","2016-10-25 01:01:36") ,("Xenos","2017-05-11 22:50:35") ,("Bert","2017-06-17 18:08:21") ,("Oleg","2018-01-03 13:10:32") ,("Hall","2018-06-04 01:53:45") ,("Evan","2017-01-16 01:04:25") ,("Mohammad","2016-11-18 05:42:52") ,("Armand","2016-12-18 06:57:57") ,("Kaseem","2018-06-12 23:09:57") ,("Colin","2017-06-29 05:25:52") ,("Arthur","2016-12-29 04:38:13") ,("Xander","2016-11-14 19:35:32") ,("Dante","2016-12-01 09:01:04") ,("Zahir","2018-02-17 14:44:53") ,("Raymond","2017-03-09 05:33:06") ,("Giacomo","2017-04-17 06:12:52") ,("Fulton","2017-06-04 00:41:57") ,("Chase","2018-01-14 03:03:57") ,("William","2017-05-08 09:44:59") ,("Fuller","2017-03-31 20:35:20") ,("Jarrod","2017-02-15 02:45:29") ,("Nissim","2018-03-11 14:19:25") ,("Chester","2017-11-05 00:14:27") ,("Perry","2017-12-24 11:58:04") ,("Theodore","2017-06-26 12:34:12") ,("Mason","2017-10-02 03:53:49") ,("Brenden","2018-10-08 10:09:47") ,("Jerome","2017-11-05 20:34:25") ,("Keaton","2018-08-18 00:55:56") ,("Tiger","2017-05-21 16:59:07") ,("Benjamin","2018-04-10 14:46:36") ,("John","2018-09-05 18:53:03") ,("Jakeem","2018-10-11 00:17:38") ,("Kenyon","2017-12-18 22:19:29") ,("Ferris","2017-03-29 06:59:13") ,("Hoyt","2017-01-03 03:48:56") ,("Fitzgerald","2017-07-27 11:27:52") ,("Forrest","2017-10-05 23:14:21") ,("Jordan","2017-01-11 03:48:09") ,("Lev","2017-05-25 08:03:39") ,("Chase","2017-06-18 19:09:23") ,("Ryder","2016-12-13 12:50:50") ,("Malik","2017-11-19 15:15:55") ,("Zeph","2018-04-04 11:22:12") ,("Amala","2017-01-29 07:52:17") ;
。
update MyTable set aNum = id ;
查询1 ：
select 'aNum by OR' q, mytable.* from mytable where aNum = 12 OR aNum = 22 OR aNum = 27 OR aNum = 32 OR aNum = 42 OR aNum = 52 OR aNum = 62 OR aNum = 65 OR aNum = 72 OR aNum = 82
结果 ：
| q | id | aName | aDate | aNum | |------------|----|----------|----------------------|------| | aNum by OR | 12 | Chadwick | 2018-08-18T08:54:08Z | 12 | | aNum by OR | 22 | Donovan | 2017-08-12T05:21:35Z | 22 | | aNum by OR | 27 | Edward | 2018-07-30T12:18:07Z | 27 | | aNum by OR | 32 | Hasad | 2017-05-15T05:32:41Z | 32 | | aNum by OR | 42 | Berk | 2017-04-16T17:39:32Z | 42 | | aNum by OR | 52 | Clayton | 2018-08-16T05:09:29Z | 52 | | aNum by OR | 62 | Mohammad | 2016-11-18T05:42:52Z | 62 | | aNum by OR | 65 | Colin | 2017-06-29T05:25:52Z | 65 | | aNum by OR | 72 | Fulton | 2017-06-04T00:41:57Z | 72 | | aNum by OR | 82 | Brenden | 2018-10-08T10:09:47Z | 82 |
查询2 ：
select 'aNum by IN' q, mytable.* from mytable where aNum IN ( 12 , 22 , 27 , 32 , 42 , 52 , 62 , 65 , 72 , 82 )
结果 ：
| q | id | aName | aDate | aNum | |------------|----|----------|----------------------|------| | aNum by IN | 12 | Chadwick | 2018-08-18T08:54:08Z | 12 | | aNum by IN | 22 | Donovan | 2017-08-12T05:21:35Z | 22 | | aNum by IN | 27 | Edward | 2018-07-30T12:18:07Z | 27 | | aNum by IN | 32 | Hasad | 2017-05-15T05:32:41Z | 32 | | aNum by IN | 42 | Berk | 2017-04-16T17:39:32Z | 42 | | aNum by IN | 52 | Clayton | 2018-08-16T05:09:29Z | 52 | | aNum by IN | 62 | Mohammad | 2016-11-18T05:42:52Z | 62 | | aNum by IN | 65 | Colin | 2017-06-29T05:25:52Z | 65 | | aNum by IN | 72 | Fulton | 2017-06-04T00:41:57Z | 72 | | aNum by IN | 82 | Brenden | 2018-10-08T10:09:47Z | 82 |
查询3 ：
select 'adate by OR' q, mytable.* from mytable where aDate= str_to_date("2017-02-15 02:45:29",'%Y-%m-%d %h:%i:%s') OR aDate = str_to_date("2018-03-10 18:37:26",'%Y-%m-%d %h:%i:%s') OR aDate = str_to_date("2017-05-20 16:25:10",'%Y-%m-%d %h:%i:%s') OR aDate = str_to_date("2018-07-10 06:29:39",'%Y-%m-%d %h:%i:%s') OR aDate = str_to_date("2017-07-14 00:04:00",'%Y-%m-%d %h:%i:%s') OR aDate = str_to_date("2018-08-18 08:54:08",'%Y-%m-%d %h:%i:%s') OR aDate = str_to_date("2018-10-02 06:55:56",'%Y-%m-%d %h:%i:%s') OR aDate = str_to_date("2017-04-20 07:03:05",'%Y-%m-%d %h:%i:%s') OR aDate = str_to_date("2018-03-02 15:04:32",'%Y-%m-%d %h:%i:%s') OR aDate = str_to_date("2017-09-19 04:07:54",'%Y-%m-%d %h:%i:%s') OR aDate = str_to_date("2016-12-13 04:24:08",'%Y-%m-%d %h:%i:%s')
结果 ：
| q | id | aName | aDate | aNum | |-------------|----|----------|----------------------|------| | adate by OR | 47 | Troy | 2016-12-13T04:24:08Z | 47 | | adate by OR | 76 | Jarrod | 2017-02-15T02:45:29Z | 76 | | adate by OR | 44 | Deacon | 2017-04-20T07:03:05Z | 44 | | adate by OR | 46 | Thaddeus | 2017-09-19T04:07:54Z | 46 | | adate by OR | 10 | Conan | 2018-07-10T06:29:39Z | 10 | | adate by OR | 12 | Chadwick | 2018-08-18T08:54:08Z | 12 | | adate by OR | 13 | Darius | 2018-10-02T06:55:56Z | 13 |
查询4 ：
select 'adate by IN' q, mytable.* from mytable where aDate IN ( str_to_date("2017-02-15 02:45:29",'%Y-%m-%d %h:%i:%s') , str_to_date("2018-03-10 18:37:26",'%Y-%m-%d %h:%i:%s') , str_to_date("2017-05-20 16:25:10",'%Y-%m-%d %h:%i:%s') , str_to_date("2018-07-10 06:29:39",'%Y-%m-%d %h:%i:%s') , str_to_date("2017-07-14 00:04:00",'%Y-%m-%d %h:%i:%s') , str_to_date("2018-08-18 08:54:08",'%Y-%m-%d %h:%i:%s') , str_to_date("2018-10-02 06:55:56",'%Y-%m-%d %h:%i:%s') , str_to_date("2017-04-20 07:03:05",'%Y-%m-%d %h:%i:%s') , str_to_date("2018-03-02 15:04:32",'%Y-%m-%d %h:%i:%s') , str_to_date("2017-09-19 04:07:54",'%Y-%m-%d %h:%i:%s') , str_to_date("2016-12-13 04:24:08",'%Y-%m-%d %h:%i:%s') )
结果 ：
| q | id | aName | aDate | aNum | |-------------|----|----------|----------------------|------| | adate by IN | 47 | Troy | 2016-12-13T04:24:08Z | 47 | | adate by IN | 76 | Jarrod | 2017-02-15T02:45:29Z | 76 | | adate by IN | 44 | Deacon | 2017-04-20T07:03:05Z | 44 | | adate by IN | 46 | Thaddeus | 2017-09-19T04:07:54Z | 46 | | adate by IN | 10 | Conan | 2018-07-10T06:29:39Z | 10 | | adate by IN | 12 | Chadwick | 2018-08-18T08:54:08Z | 12 | | adate by IN | 13 | Darius | 2018-10-02T06:55:56Z | 13 |
查询5 ：
select 'name by OR' q, mytable.* from mytable where aname = 'Alan' OR aname = 'Brian' OR aname = 'Chandler' OR aname = 'Darius' OR aname = 'Evan' OR aname = 'Ferris' OR aname = 'Giacomo' OR aname = 'Hall' OR aname = 'James' OR aname = 'Jarrod'
结果 ：
| q | id | aName | aDate | aNum | |-------------|----|----------|----------------------|------| | name by OR | 43 | Alan | 2018-09-01T05:33:05Z | 43 | | name by OR | 24 | Brian | 2017-02-01T23:35:20Z | 24 | | name by OR | 26 | Chandler | 2017-09-30T23:54:06Z | 26 | | name by OR | 13 | Darius | 2018-10-02T06:55:56Z | 13 | | name by OR | 61 | Evan | 2017-01-16T01:04:25Z | 61 | | name by OR | 90 | Ferris | 2017-03-29T06:59:13Z | 90 | | name by OR | 37 | Giacomo | 2018-10-06T14:21:21Z | 37 | | name by OR | 71 | Giacomo | 2017-04-17T06:12:52Z | 71 | | name by OR | 16 | Hall | 2017-10-13T00:17:24Z | 16 | | name by OR | 60 | Hall | 2018-06-04T01:53:45Z | 60 | | name by OR | 38 | James | 2018-02-06T02:45:59Z | 38 | | name by OR | 76 | Jarrod | 2017-02-15T02:45:29Z | 76 |
查询6 ：
select 'name by IN' q, mytable.* from mytable where aname IN ( 'Alan' ,'Brian' ,'Chandler' , 'Darius' , 'Evan' , 'Ferris' , 'Giacomo' , 'Hall' , 'James' , 'Jarrod' )
结果 ：
| q | id | aName | aDate | aNum | |------------|----|----------|----------------------|------| | name by IN | 43 | Alan | 2018-09-01T05:33:05Z | 43 | | name by IN | 24 | Brian | 2017-02-01T23:35:20Z | 24 | | name by IN | 26 | Chandler | 2017-09-30T23:54:06Z | 26 | | name by IN | 13 | Darius | 2018-10-02T06:55:56Z | 13 | | name by IN | 61 | Evan | 2017-01-16T01:04:25Z | 61 | | name by IN | 90 | Ferris | 2017-03-29T06:59:13Z | 90 | | name by IN | 37 | Giacomo | 2018-10-06T14:21:21Z | 37 | | name by IN | 71 | Giacomo | 2017-04-17T06:12:52Z | 71 | | name by IN | 16 | Hall | 2017-10-13T00:17:24Z | 16 | | name by IN | 60 | Hall | 2018-06-04T01:53:45Z | 60 | | name by IN | 38 | James | 2018-02-06T02:45:59Z | 38 | | name by IN | 76 | Jarrod | 2017-02-15T02:45:29Z | 76 |
