HIVE中EXPLAIN的解读方式

1.执行计划可以知道关键字的执行先后顺序（这里与oracle的会有些不同）

首先来执行一个很简单的sql语句。

hive (dwd)> explain select level,gender,count(*) from user_behavior where channel = 'huawei' group by level,gender;
OK
Explain
STAGE DEPENDENCIES:*--表明当前stage所有的依赖顺序*
  Stage-1 is a root stage
  Stage-0 depends on stages: Stage-1
*--从上文可以看出stage-0 依赖于stage-1，会等待stage-1执行完毕，stage-0才会执行。*
STAGE PLANS:
  Stage: Stage-1*--表明当前显示的是stage-1的执行计划，属于整体计划的子计划。*
    Tez*--当时用的tez引擎，没留意*
      Edges:
        Reducer 2 <- Map 1 (SIMPLE_EDGE)
      DagName: atguigu_20190814062239_a68f9d23-6b82-4d39-b9fd-53704c3382e1:3
      Vertices:
        Map 1 
            Map Operator Tree:
                TableScan*--先扫描表*
                  alias: user_behavior*--表名*
                  Statistics: Num rows: 620000 Data size: 580674000 Basic stats: COMPLETE Column stats: NONE
                  Filter Operator
                    predicate: (channel = 'huawei') (type: boolean)*--而后过滤*
                    Statistics: Num rows: 310000 Data size: 290337000 Basic stats: COMPLETE Column stats: NONE
                    Select Operator
                      expressions: level (type: tinyint), gender (type: string)*--继而select2个字段*
                      outputColumnNames: level, gender
                      Statistics: Num rows: 310000 Data size: 290337000 Basic stats: COMPLETE Column stats: NONE
                      Group By Operator
                        aggregations: count()
                        keys: level (type: tinyint), gender (type: string)*--group by做聚合操作*
                        mode: hash
                        outputColumnNames: _col0, _col1, _col2
                        Statistics: Num rows: 310000 Data size: 290337000 Basic stats: COMPLETE Column stats: NONE
                        Reduce Output Operator
                          key expressions: _col0 (type: tinyint), _col1 (type: string)*--order by默认升序排序，虽然我们sql中没写，但是会默认做*
                          sort order: ++
                          Map-reduce partition columns: _col0 (type: tinyint), _col1 (type: string)
                          Statistics: Num rows: 310000 Data size: 290337000 Basic stats: COMPLETE Column stats: NONE
                          value expressions: _col2 (type: bigint)
        Reducer 2 
            Reduce Operator Tree:
              Group By Operator
                aggregations: count(VALUE._col0)
                keys: KEY._col0 (type: tinyint), KEY._col1 (type: string)*--group by做聚合操作，刚才的map1也有同样的操作，推测是预聚合。这里才是最终结果。*
                mode: mergepartial
                outputColumnNames: _col0, _col1, _col2
                Statistics: Num rows: 155000 Data size: 145168500 Basic stats: COMPLETE Column stats: NONE
                File Output Operator
                  compressed: false*--压缩方式没有，false*
                  Statistics: Num rows: 155000 Data size: 145168500 Basic stats: COMPLETE Column stats: NONE
                  table:
                      input format: org.apache.hadoop.mapred.TextInputFormat
                      output format: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat
                      serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe
*--至此stage-1结束。接下来进入stage-0*
  Stage: Stage-0
    Fetch Operator
      limit: -1