MySQL INNER JOIN算法的效率分析

MySQL处理JOIN的方法如下：（摘自MySQL 5.1 参考手册中文版）

假定我们有一个如下形式的表T1、T2、T3的联接查询：

SELECT * FROM T1 INNER JOIN T2 ON P1(T1,T2) INNER JOIN T3 ON P2(T2,T3) WHERE P(T1,T2,T3).

这里，P1(T1,T2)和P2(T3,T3)是一些联接条件(表达式)，其中P(t1,t2,t3)是表T1、T2、T3的列的一个条件。

嵌套环联接算法将按下面的方式执行该查询：

FOR each row t1 in T1 { FOR each row t2 in T2 such that P1(t1,t2) { FOR each row t3 in T3 such that P2(t2,t3) { IF P(t1,t2,t3) { t:=t1||t2||t3; OUTPUT t; } } } }

符号t1||t2||t3表示“连接行t1、t2和t3的列组成的行”。

其实我觉得，完全可以把P(t1, t2, t3)拆到进入循环前就处理，像这样（后来仔细看了文档，MySQL在内联接的时候还是会优化成这样的）：

FOR each row t1 in T1 { IF P(t1) { FOR each row t2 in T2 such that P1(t1,t2) { IF P(t2) { FOR each row t3 in T3 such that P2(t2,t3) { IF P(t3) { t:=t1||t2||t3; OUTPUT t; } } } } } }

甚至更快的是把条件全部合并起来：

FOR each row t1 in T1 { IF P(t1) { FOR each row t2 in T2 such that (P1(t1,t2) && P(t2)) { FOR each row t3 in T3 such that (P2(t2,t3) && P(t3)) { t:=t1||t2||t3; OUTPUT t; } } } }

我写了个程序，把方法一（MySQL的方法）和方法三的效率进行比较，明显方法三要高。

#include #include #include #define MAXN 100000   using namespace std;   int a[MAXN]; int b[MAXN]; int c[MAXN]; int count = 0;   int main() { clock_t start, finish; double time1, time2;   count = 0; srand(time(0));   for(int i=0; i<MAXN; ++i) { a[i] = i; b[i] = MAXN-i; c[i] = rand()%MAXN; }   start=clock(); for(int i=0; i<MAXN; ++i) { for(int j=10; j<MAXN; ++j) { if (a[i]==b[j]) { for(int k=0; k<MAXN; ++k) { if(b[j]==c[k]) { if (a[i]>500 and b[j] < 800 and c[k]>120) { cout << ++count << ':' <<a[i] << ',' << b[j] << ',' << c[k] << endl; } } } } } } finish = clock(); time1 = (double)(finish-start)/CLOCKS_PER_SEC;   count = 0; start=clock(); for(int i=0; i<MAXN; ++i) { if (a[i]>500) { for(int j=10; j<MAXN; ++j) { if (a[i]==b[j] and b[j] < 800) { for(int k=0; k<MAXN; ++k) { if(b[j]==c[k] and c[k]>120) { cout << ++count << ':' <<a[i] << ',' << b[j] << ',' << c[k] << endl; } } } } } } finish = clock(); time2 = (double)(finish-start)/CLOCKS_PER_SEC;   cout << time1 << "VS" << time2 << endl; return 0; }