impala数据倾斜 impala 偏移条款

一般来说，select查询的resultset中的行从0开始。使用offset子句，我们可以决定从哪里考虑输出。 例如，如果我们选择偏移为0，结果将像往常一样，如果我们选择偏移为5，结果从第五行开始。

语法

以下是Impala中的biasclause的语法。

select data from table_name Group BY col_name;

例

假设我们在数据库my_db中有一个名为customers的表，其内容如下 -

[quickstart.cloudera:21000] > select * from customers; 
Query: select * from customers 
+----+----------+-----+-----------+--------+ 
| id | name     | age | address   | salary | 
+----+----------+-----+-----------+--------+ 
| 3  | kaushik  | 23  | Kota      | 30000  | 
| 6  | Komal    | 22  | MP        | 32000  | 
| 1  | Ramesh   | 32  | Ahmedabad | 20000  | 
| 5  | Hardik   | 27  | Bhopal    | 40000  | 
| 2  | Khilan   | 25  | Delhi     | 15000  | 
| 8  | ram      | 22  | vizag     | 31000  | 
| 9  | robert   | 23  | banglore  | 28000  |
| 7  | ram      | 25  | chennai   | 23000  | 
| 4  | Chaitali | 25  | Mumbai    | 35000  | 
+----+----------+-----+-----------+--------+ 
Fetched 9 row(s) in 0.51s

您可以按其id的升序排列表中的记录，并使用limit和order by子句将记录数限制为4，如下所示。

Query: select * from customers order by id limit 4 
+----+----------+-----+-----------+--------+ 
| id | name     | age | address   | salary | 
+----+----------+-----+-----------+--------+ 
| 1  | Ramesh   | 32  | Ahmedabad | 20000  | 
| 2  | Khilan   | 25  | Delhi     | 15000  | 
| 3  | kaushik  | 23  | Kota      | 30000  | 
| 4  | Chaitali | 25  | Mumbai    | 35000  | 
+----+----------+-----+-----------+--------+ 
Fetched 4 row(s) in 0.64s

以下是偏移子句的示例。 这里，我们按照id的顺序在customers表中获取记录，并从第0行开始打印前四行。

[quickstart.cloudera:21000] > select * from customers order by id limit 4 offset 0;

执行时，上述查询给出以下结果。

Query: select * from customers order by id limit 4 offset 0 
+----+----------+-----+-----------+--------+
| id | name     | age | address   | salary | 
+----+----------+-----+-----------+--------+ 
| 1  | Ramesh   | 32  | Ahmedabad | 20000  | 
| 2  | Khilan   | 25  | Delhi     | 15000  | 
| 3  | kaushik  | 23  | Kota      | 30000  | 
| 4  | Chaitali | 25  | Mumbai    | 35000  | 
+----+----------+-----+-----------+--------+ 
Fetched 4 row(s) in 0.62s

以相同的方式，您可以从具有偏移5的行开始从客户表获取四个记录，如下所示。

[quickstart.cloudera:21000] > select * from customers order by id limit 4 offset 5; 
Query: select * from customers order by id limit 4 offset 5 
+----+--------+-----+----------+--------+ 
| id | name   | age | address  | salary | 
+----+--------+-----+----------+--------+ 
| 6  | Komal  | 22  | MP       | 32000  | 
| 7  | ram    | 25  | chennai  | 23000  | 
| 8  | ram    | 22  | vizag    | 31000  |
| 9  | robert | 23  | banglore | 28000  | 
+----+--------+-----+----------+--------+ 
Fetched 4 row(s) in 0.52s