SQL關(guān)聯(lián)查詢 直接join 和子查詢的區(qū)別

運(yùn)營組的同事最近提出一個需求，希望可以統(tǒng)計(jì)出用系統(tǒng)用戶及訂單情況，于是乎我們很想當(dāng)然的寫出了一個統(tǒng)計(jì)SQL，用戶表user和行程表直接join，并且針對行程做了group，但SQL執(zhí)行速度出奇的慢。

explain select  users.`mobile_num`, concat(users.`lastName` ,users.`firstName`) as userName, users.`company`,
  (case `users`.`idPhotoCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `idPhotoCheckStatus`,
  (case `users`.`driverLicenseCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `driverLicenseCheckStatus`,
  (case `users`.`companyCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `companyCheckStatus`,
  (case `users`.`unionCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `unionCheckStatus`,
  count(passenger_trip.id) as ptrip_num
from users
left join passenger_trip on passenger_trip.userId = users.id  and passenger_trip.status != 'cancel'
left join driver_trip on driver_trip.`userId`=users.`id` and driver_trip.`status` != 'cancel'
where company != '本公司名' and company != '本公司昵稱'

當(dāng)時的第一反應(yīng)是數(shù)據(jù)庫掛住了，因?yàn)橛脩舯淼臄?shù)據(jù)量10W左右，行程表的數(shù)據(jù)也是10W左右，不可能這么慢！通過explain查看分析計(jì)劃，并且查看過關(guān)聯(lián)字段的索引情況，發(fā)現(xiàn)這是一個最常見的關(guān)聯(lián)查詢，當(dāng)然是通過join實(shí)現(xiàn)。

轉(zhuǎn)而一想，10W*10W，經(jīng)過笛卡爾集之后，這不是百億級的數(shù)據(jù)篩選嗎？！于是換了一種寫法進(jìn)行嘗試。

explain select  users.`mobile_num`, concat(users.`lastName` ,users.`firstName`) as userName, users.`company`,
  (case `users`.`idPhotoCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `idPhotoCheckStatus`,
  (case `users`.`driverLicenseCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `driverLicenseCheckStatus`,
  (case `users`.`companyCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `companyCheckStatus`,
  (case `users`.`unionCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `unionCheckStatus`,
  (select count(passenger_trip.id) from  passenger_trip where  passenger_trip.userId = users.id  and passenger_trip.status != 'cancel') as ptrip_num,
  (select count(driver_trip.id) from  driver_trip where  driver_trip.userId = users.id  and driver_trip.status != 'cancel') as dtrip_num
from users
where company != '本公司名' and company != '公司昵稱'

這樣的效果居然比直接join快了N倍，執(zhí)行速度從未知到10秒內(nèi)返回，查看執(zhí)行計(jì)劃：

進(jìn)一步調(diào)整SQL進(jìn)行嘗試：

explain select  users.`mobile_num`, concat(users.`lastName` ,users.`firstName`) as userName, users.`company`,
  (case `users`.`idPhotoCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `idPhotoCheckStatus`,
  (case `users`.`driverLicenseCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `driverLicenseCheckStatus`,
  (case `users`.`companyCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `companyCheckStatus`,
  (case `users`.`unionCheckStatus` when '2' then '已認(rèn)證' when '3' then '已駁回' else '待認(rèn)證' end) as `unionCheckStatus`,
 ptrip_num, dtrip_num
from users 
 left  join 
 (select count(passenger_trip.id)  as ptrip_num, passenger_trip.`userId` from  passenger_trip where  passenger_trip.status != 'cancel' group by passenger_trip.`userId` ) as ptrip
 on ptrip.userId = users.id
 left join 
 (select count(driver_trip.id)  as dtrip_num, driver_trip.`userId` from  driver_trip where  driver_trip.status != 'cancel' group by driver_trip.`userId` ) as dtrip
 on dtrip.userId = users.id
where company != '本公司名' and company != '公司昵稱'

居然5秒內(nèi)返回，這才是正常的預(yù)期，10W級的數(shù)據(jù)篩選，應(yīng)該是幾秒內(nèi)返回的！

出現(xiàn)這種差別的原因，其實(shí)很簡單，SQL語句執(zhí)行的時候是有一定順序的。

1. from 先選擇一個表，構(gòu)成一個結(jié)果集。
2. where 對結(jié)果集進(jìn)行篩選，篩選出需要的信息形成新的結(jié)果集。
3. group by 對新的結(jié)果集分組。
4. having 篩選出想要的分組。
5. select 選擇列。
6. order by 當(dāng)所有的條件都弄完了。最后排序。

第一種寫法，直接join的結(jié)果，就是在100億條數(shù)據(jù)中進(jìn)行篩選；
后面兩種則是優(yōu)先執(zhí)行子查詢，完成10W級別的查詢，再進(jìn)行一次主表10W級的關(guān)聯(lián)查詢，所以數(shù)量級明顯少于第一種寫法。