Question

所以有两个查询。为什么第一次运行速度比第二次快？如何优化内连接的速度与第一个一样快？

总记录超过4 000 000

idTerminals已编入索引


select (select name 
          from terminals 
          where terminals.idTerminals = AnalyzeProfitResults.idTerminals),
        currency, 
        count(*) 
from AnalyzeProfitResults
group by AnalyzeProfitResults.idTerminals,currency;


select name,
        currency, 
        count(*) 
from AnalyzeProfitResults
inner join terminals on terminals.idTerminals = AnalyzeProfitResults.idTerminals
group by AnalyzeProfitResults.idTerminals,currency;

结果相同：


name,currency,count(*)
"Buy or sell",EURUSD,235105
"Buy or sell",GBPUSD,14515
"Buy or sell",USDJPY,235298
"very strict",AUDJPY,234787
"very strict",AUDUSD,235298
"very strict",CHFJPY,235298
"very strict",EURJPY,235298
"very strict",EURUSD,235298
"very strict",GBPJPY,235298
"very strict",GBPUSD,235298
"very strict",NZDUSD,235298
"very strict",USDCHF,235298
"very strict",USDJPY,235298
",normal",AUDJPY,235298
",normal",AUDUSD,36391
",normal",EURUSD,234985
",normal",GBPJPY,117649
",normal",NZDUSD,198552
",normal",USDCHF,235298
",normal",USDJPY,235298

Answer 1

我想在后一个查询中，MySQL加入AnalyzeProfitResults和terminals，为size(AnalyzeProfitResults) * size(terminals)操作产生group by行。在AnalyzeProfitResults中有400万行，即使terminals只包含少量记录，也会有很多行。在前一种情况下，MySQL在400万行上执行分组，产生你给出的20行行，然后只对20行中的每一行执行子查询（甚至可以在内部重写为连接，这将是偶数快）。

修改这是第一个带有显式连接的查询：

select terminals.name, APR_aggregated.* from terminals join ( select APR.currency, count(*) from AnalyzeProfitResults as APR group by APR.idTerminals, APR.currency ) as APR_aggregated on terminals.idTerminals = APR_aggregated.idTerminals

Mysql优化：有人可以解释为什么一个查询比内连接更快

1 个答案: