mirror of
https://github.com/ClickHouse/ClickHouse.git
synced 2024-09-25 03:00:49 +00:00
d91c97d15d
* Replace underscores with hyphens * remove temporary code * fix style check * fix collapse
95 lines
3.3 KiB
Markdown
95 lines
3.3 KiB
Markdown
# AggregatingMergeTree {#aggregatingmergetree}
|
||
|
||
该引擎继承自 [MergeTree](mergetree.md)，并改变了数据片段的合并逻辑。 ClickHouse 会将相同主键的所有行（在一个数据片段内）替换为单个存储一系列聚合函数状态的行。
|
||
|
||
可以使用 `AggregatingMergeTree` 表来做增量数据统计聚合，包括物化视图的数据聚合。
|
||
|
||
引擎需使用 [AggregateFunction](../../../engines/table-engines/mergetree-family/aggregatingmergetree.md) 类型来处理所有列。
|
||
|
||
如果要按一组规则来合并减少行数，则使用 `AggregatingMergeTree` 是合适的。
|
||
|
||
## 建表 {#jian-biao}
|
||
|
||
``` sql
|
||
CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster]
|
||
(
|
||
name1 [type1] [DEFAULT|MATERIALIZED|ALIAS expr1],
|
||
name2 [type2] [DEFAULT|MATERIALIZED|ALIAS expr2],
|
||
...
|
||
) ENGINE = AggregatingMergeTree()
|
||
[PARTITION BY expr]
|
||
[ORDER BY expr]
|
||
[SAMPLE BY expr]
|
||
[SETTINGS name=value, ...]
|
||
```
|
||
|
||
语句参数的说明，请参阅 [语句描述](../../../engines/table-engines/mergetree-family/aggregatingmergetree.md)。
|
||
|
||
**子句**
|
||
|
||
创建 `AggregatingMergeTree` 表时，需用跟创建 `MergeTree` 表一样的[子句](mergetree.md)。
|
||
|
||
<details markdown="1">
|
||
|
||
<summary>已弃用的建表方法</summary>
|
||
|
||
!!! 注意 "注意"
|
||
不要在新项目中使用该方法，可能的话，请将旧项目切换到上述方法。
|
||
|
||
``` sql
|
||
CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster]
|
||
(
|
||
name1 [type1] [DEFAULT|MATERIALIZED|ALIAS expr1],
|
||
name2 [type2] [DEFAULT|MATERIALIZED|ALIAS expr2],
|
||
...
|
||
) ENGINE [=] AggregatingMergeTree(date-column [, sampling_expression], (primary, key), index_granularity)
|
||
```
|
||
|
||
上面的所有参数跟 `MergeTree` 中的一样。
|
||
</details>
|
||
|
||
## SELECT 和 INSERT {#select-he-insert}
|
||
|
||
插入数据，需使用带有聚合 -State- 函数的 [INSERT SELECT](../../../engines/table-engines/mergetree-family/aggregatingmergetree.md) 语句。
|
||
从 `AggregatingMergeTree` 表中查询数据时，需使用 `GROUP BY` 子句并且要使用与插入时相同的聚合函数，但后缀要改为 `-Merge` 。
|
||
|
||
在 `SELECT` 查询的结果中，对于 ClickHouse 的所有输出格式 `AggregateFunction` 类型的值都实现了特定的二进制表示法。如果直接用 `SELECT` 导出这些数据，例如如用 `TabSeparated` 格式，那么这些导出数据也能直接用 `INSERT` 语句加载导入。
|
||
|
||
## 聚合物化视图的示例 {#ju-he-wu-hua-shi-tu-de-shi-li}
|
||
|
||
创建一个跟踪 `test.visits` 表的 `AggregatingMergeTree` 物化视图：
|
||
|
||
``` sql
|
||
CREATE MATERIALIZED VIEW test.basic
|
||
ENGINE = AggregatingMergeTree() PARTITION BY toYYYYMM(StartDate) ORDER BY (CounterID, StartDate)
|
||
AS SELECT
|
||
CounterID,
|
||
StartDate,
|
||
sumState(Sign) AS Visits,
|
||
uniqState(UserID) AS Users
|
||
FROM test.visits
|
||
GROUP BY CounterID, StartDate;
|
||
```
|
||
|
||
向 `test.visits` 表中插入数据。
|
||
|
||
``` sql
|
||
INSERT INTO test.visits ...
|
||
```
|
||
|
||
数据会同时插入到表和视图中，并且视图 `test.basic` 会将里面的数据聚合。
|
||
|
||
要获取聚合数据，我们需要在 `test.basic` 视图上执行类似 `SELECT ... GROUP BY ...` 这样的查询 ：
|
||
|
||
``` sql
|
||
SELECT
|
||
StartDate,
|
||
sumMerge(Visits) AS Visits,
|
||
uniqMerge(Users) AS Users
|
||
FROM test.basic
|
||
GROUP BY StartDate
|
||
ORDER BY StartDate;
|
||
```
|
||
|
||
[来源文章](https://clickhouse.tech/docs/en/operations/table_engines/aggregatingmergetree/) <!--hide-->
|