Flink SQL之Over 聚合操作

Over 聚合定义（支持 Batch\Streaming）：可以理解为是一种特殊的滑动窗口聚合函数。那这里我们拿 Over 聚合​ 与 窗口聚合 做一个对比，其之间的最大不同之处在于：窗口聚合：不在 group by 中的字段，不能直接在 select 中拿到;Over 聚合：能够保留原始字段.在生产环境中，Over 聚合的使用场景还是比较少的。在 Hive 中也有相同的聚合，但是小伙伴萌可以想想你在离线数仓经常使用嘛？

应用场景：计算最近一段滑动窗口的聚合结果数据。
实际案例：查询每个产品最近一小时订单的金额总和：

SELECT order_id, order_time, amount,
  SUM(amount) OVER (
    PARTITION BY product
    ORDER BY order_time
    RANGE BETWEEN INTERVAL '1' HOUR PRECEDING AND CURRENT ROW
  ) AS one_hour_prod_amount_sum
FROM Orders

Over 聚合的语法总结如下：

SELECT
  agg_func(agg_col) OVER (
    [PARTITION BY col1[, col2, ...]]
    ORDER BY time_col
    range_definition),
  ...
FROM ...

其中：

• ORDER BY：必须是时间戳列（事件时间、处理时间）
• PARTITION BY：标识了聚合窗口的聚合粒度，如上述案例是按照 product 进行聚合
• range_definition：这个标识聚合窗口的聚合数据范围，在 Flink 中有两种指定数据范围的方式。第一种为按照行数聚合​，第二种为按照时间区间聚合。

如下案例所示：

时间区间聚合

按照时间区间聚合就是时间区间的一个滑动窗口，比如下面案例 1 小时的区间，最新输出的一条数据的 sum 聚合结果就是最近一小时数据的 amount 之和。

CREATE TABLE source_table (
    order_id BIGINT,
    product BIGINT,
    amount BIGINT,
    order_time as cast(CURRENT_TIMESTAMP as TIMESTAMP(3)),
    WATERMARK FOR order_time AS order_time - INTERVAL '0.001' SECOND
) WITH (
  'connector' = 'datagen',
  'rows-per-second' = '1',
  'fields.order_id.min' = '1',
  'fields.order_id.max' = '2',
  'fields.amount.min' = '1',
  'fields.amount.max' = '10',
  'fields.product.min' = '1',
  'fields.product.max' = '2'
);

CREATE TABLE sink_table (
    product BIGINT,
    order_time TIMESTAMP(3),
    amount BIGINT,
    one_hour_prod_amount_sum BIGINT
) WITH (
  'connector' = 'print'
);

INSERT INTO sink_table
SELECT product, order_time, amount,
  SUM(amount) OVER (
    PARTITION BY product
    ORDER BY order_time
    -- 标识统计范围是一个 product 的最近 1 小时的数据
    RANGE BETWEEN INTERVAL '1' HOUR PRECEDING AND CURRENT ROW
  ) AS one_hour_prod_amount_sum
FROM source_table

结果如下：

+I[2, 2021-12-24T22:08:26.583, 7, 73]
+I[2, 2021-12-24T22:08:27.583, 7, 80]
+I[2, 2021-12-24T22:08:28.583, 4, 84]
+I[2, 2021-12-24T22:08:29.584, 7, 91]
+I[2, 2021-12-24T22:08:30.583, 8, 99]
+I[1, 2021-12-24T22:08:31.583, 9, 138]
+I[2, 2021-12-24T22:08:32.584, 6, 105]
+I[1, 2021-12-24T22:08:33.584, 7, 145]

行数聚合

按照行数聚合就是数据行数的一个滑动窗口，比如下面案例，最新输出的一条数据的 sum 聚合结果就是最近 5 行数据的 amount 之和。

CREATE TABLE source_table (
    order_id BIGINT,
    product BIGINT,
    amount BIGINT,
    order_time as cast(CURRENT_TIMESTAMP as TIMESTAMP(3)),
    WATERMARK FOR order_time AS order_time - INTERVAL '0.001' SECOND
) WITH (
  'connector' = 'datagen',
  'rows-per-second' = '1',
  'fields.order_id.min' = '1',
  'fields.order_id.max' = '2',
  'fields.amount.min' = '1',
  'fields.amount.max' = '2',
  'fields.product.min' = '1',
  'fields.product.max' = '2'
);

CREATE TABLE sink_table (
    product BIGINT,
    order_time TIMESTAMP(3),
    amount BIGINT,
    one_hour_prod_amount_sum BIGINT
) WITH (
  'connector' = 'print'
);

INSERT INTO sink_table
SELECT product, order_time, amount,
  SUM(amount) OVER (
    PARTITION BY product
    ORDER BY order_time
    -- 标识统计范围是一个 product 的最近 5 行数据
    ROWS BETWEEN 5 PRECEDING AND CURRENT ROW
  ) AS one_hour_prod_amount_sum
FROM source_table

预跑结果如下：

+I[2, 2021-12-24T22:18:19.147, 1, 9]
+I[1, 2021-12-24T22:18:20.147, 2, 11]
+I[1, 2021-12-24T22:18:21.147, 2, 12]
+I[1, 2021-12-24T22:18:22.147, 2, 12]
+I[1, 2021-12-24T22:18:23.148, 2, 12]
+I[1, 2021-12-24T22:18:24.147, 1, 11]
+I[1, 2021-12-24T22:18:25.146, 1, 10]
+I[1, 2021-12-24T22:18:26.147, 1, 9]
+I[2, 2021-12-24T22:18:27.145, 2, 11]
+I[2, 2021-12-24T22:18:28.148, 1, 10]
+I[2, 2021-12-24T22:18:29.145, 2, 10]

当然，如果你在一个 SELECT 中有多个聚合窗口的聚合方式，Flink SQL 支持了一种简化写法，如下案例：

SELECT order_id, order_time, amount,
  SUM(amount) OVER w AS sum_amount,
  AVG(amount) OVER w AS avg_amount
FROM Orders
-- 使用下面子句，定义 Over Window
WINDOW w AS (
  PARTITION BY product
  ORDER BY order_time
  RANGE BETWEEN INTERVAL '1' HOUR PRECEDING AND CURRENT ROW)