左连接 - 如何更高效？

我写了一个难以阅读的查询，我担心效率可能不好。根据从每个表到consumerdata.consumerid的消息匹配，可以更有效地编写此查询吗？左连接 - 如何更高效？

select 

consumerdata.consumerid, 
signupdate, 
city, 
state, 
year(DOB), 
topaffiliate, 
activestatus, 
lastuseddate, 

(select sum(achload.transactionamount) from achload where achload.consumerid = consumerdata.consumerid) as Total_ACH, 
(select sum(billpay.transactionamount) from billpay where billpay.consumerid = consumerdata.consumerid) as Bill_Pay, 
(select sum(recharge.transactionamount) from recharge where recharge.consumerid = consumerdata.consumerid) as Revenue, 
(select count(cash.consumerid) from Cash where cash.consumerid = consumerdata.consumerid) as Cash__TXNs, 
(select sum(moneytransfer.transactionamount) from moneytransfer where moneytransfer.consumerid = consumerdata.consumerid) as Transfer, 
(select sum(moneytransfer.commissionfeeamount) from moneytransfer where moneytransfer.consumerid = tbl_accounts_consumerdata.consumerid) as commission_fee, 
(select count(interchangetransactions.transactiondescription) from interchangetransactions where interchangetransactions.transactiondescription = "Withdrawal" and interchangetransactions.consumerid = consumerdata.consumerid) as Withdrawals, 
(select count(interchangetransactions.responsecode) from interchangetransactions where interchangetransactions.responsecode in (01) and interchangetransactions.consumerid = consumerdata.consumerid and interchangetransactions.transactiondescription = "Withdrawal") as Code_01, 
(select count(interchangetransactions.responsecode) from interchangetransactions where interchangetransactions.responsecode in (02) and interchangetransactions.consumerid = consumerdata.consumerid and interchangetransactions.transactiondescription = "Withdrawal") as Code_02, 
(select count(interchangetransactions.transactiondescription) from interchangetransactions where interchangetransactions.transactiondescription IN ("2 Cash Advance", "1 Cash Advance") and interchangetransactions.consumerid = consumerdata.consumerid) as Cash_Advance_Count, 
(select count(interchangetransactions.transactiondescription) from interchangetransactions where interchangetransactions.transactiondescription IN ("Balance") and interchangetransactions.consumerid = consumerdata.consumerid) as Inquiry_Count, 
(select sum(interchangetransactions.transactionamount) from interchangetransactions where interchangetransactions.transactiontypecode = 195 and interchangetransactions.consumerid = consumerdata.consumerid) as Intnl_TXN_Sum, 
(select count(interchangetransactions.transactionamount) from interchangetransactions where interchangetransactions.transactiontypecode = 195 and interchangetransactions.consumerid = consumerdata.consumerid) as Intnl_TXN_Count 


from 
consumerdata 

left join achload 
on achload.consumerid=consumerdata.consumerid 

left join billpay 
on billpay.consumerid = consumerdata.consumerid 

left join recharge 
on recharge.consumerid = consumerdata.consumerid 

left join cash 
on cash.consumerid = consumerdata.consumerid 

left join moneytransfer 
on moneytransfer.consumerid = consumerdata.consumerid 

left join interchangetransactions 
on interchangetransactions.consumerid = consumerdata.consumerid 

where consumerdata.signupdate between date(20120101) and curdate() 
group by consumerdata.consumerid order by signupdate asc;

来源

2016-09-22 Edward Ko

对于在子查询中查询的同一表，您还有'LEFT JOIN'似乎很奇怪。你不应该两个都做。 – Barmar

哪些表做其他'SELECT'列是从哪里来的？ – Barmar

顶部的select语句都来自consumerdata表。如果你指的是城市，州，年（dob）。 –

既然你不选择从任何交易表的东西，简单的改善将是摆脱所有LEFT JOIN条款。他们没有被用于任何事情，因为你从相关的子查询中获得总数。这导致数据库在所有表之间创建一个巨大的交叉产品，然后在底部执行GROUP BY时将其丢弃。

另一种方法是使用子查询来计算每个事务表的分组总数，而不是相关子查询。

select 
    c.consumerid, signupdate, state, year(DOB), topaffiliate, activestatus, lastuseddate, 
    IFNULL(Total_ACH, 0) AS Total_ACH, IFNULL(Bill_Pay, 0) AS Bill_Pay, IFNULL(Revenue, 0) AS Revenue, 
    IFNULL(Cash__TXNs, 0) AS Cash__TXNs, ... 
FROM consumerdata AS c 
LEFT JOIN (
    SELECT consumerid, SUM(transactionamount) AS Total_ACH 
    FROM achload 
    GROUP BY consumerid) AS a ON a.consumerid = c.consumerid 
LEFT JOIN (
    SELECT consumerid, SUM(transactionamount) AS Bill_Pay 
    FROM billpay 
    GROUP BY consumerid) AS b ON b.consumerid = c.consumerid 
LEFT JOIN (
    SELECT consumerid, SUM(transactionamount) AS Revenue 
    FROM recharge 
    GROUP BY consumerid) AS r ON r.consumerid = c.consumerid 
LEFT JOIN (
    SELECT consumerid, COUNT(*) AS Cash__TXNs 
    FROM Cash 
    GROUP BY consumerid) AS cash ON cash.consumerid = c.consumerid 
... 
WHERE c.signupdate between DATE(20120101) and CURDATE() 
ORDER BY signupdate ASC

我不知道哪条路实际上是更快 - 它很可能将取决于有多少数据是在对消费者，其注册日期指定范围之外的事务表。试试看看哪个更好。

来源

2016-09-22 16:27:17 Barmar

非常感谢你的帮助Barmar。查看你的答案后，我现在对如何利用左连接有了更好的理解。 –

左连接 - 如何更高效？

回答

相关问题