EXPLAIN 输出格式

输出结构

语法结构执行结果是为 statement 选择的计划的文本描述，可以选择使用执行统计信息进行注释。

以下以 SQL 为例，演示输出结构：

explain select city,libname1,count(libname1) as a from t3 join t1 on libname1=libname3 join t2 on isbn3=isbn2 group by city,libname1;

+--------------------------------------------------------------------------------------------+
| QUERY PLAN                                                                                 |
+--------------------------------------------------------------------------------------------+
| Project(cost=0.00..0.00 card=400.00 ndv=0.00 rowsize=0                                     |
|   ->  Aggregate(cost=0.00..0.00 card=400.00 ndv=0.00 rowsize=0                             |
|         Group Key:#[0,1], #[0,0]                                                           |
|         Aggregate Functions: count(#[0,0])                                                 |
|         ->  Join(cost=0.00..0.00 card=400.00 ndv=0.00 rowsize=0                            |
|               Join Type: INNER                                                             |
|               Join Cond: (#[1,2] = #[0,0])                                                 |
|               ->  Table Scan on abc.t2(cost=0.00..0.00 card=8.00 ndv=0.00 rowsize=0        |
|               ->  Join(cost=0.00..0.00 card=50.00 ndv=0.00 rowsize=0                       |
|                     Join Type: INNER                                                       |
|                     Join Cond: (#[0,0] = #[1,1])                                           |
|                     ->  Table Scan on abc.t1(cost=0.00..0.00 card=5.00 ndv=0.00 rowsize=0  |
|                     ->  Table Scan on abc.t3(cost=0.00..0.00 card=10.00 ndv=0.00 rowsize=0 |
+--------------------------------------------------------------------------------------------+
13 rows in set (0.00 sec)

EXPLAIN 输出一个名称为 Execution Plan Tree 树形结构，每个叶子节点都包含节点类型、受影响的对象以及其他属性的信息，如 cost，rowsize 等。我们现在只使用节点类型信息来简化展示上面的示例。Execution Plan Tree 树形结构可以可视化 SQL 查询的整个过程，显示它所经过的操作节点以及它们的成本估计。

Project
└── Aggregate
    └── Join
        └── Table Scan
        └── Join
              └──Table Scan
              └──Table Scan

节点类型

MatrixOne 支持以下节点类型。

Table Scan

Values Scan

Project

Aggregate

Filter

Join

Sort