Understanding Flink

Flink 下载：

bash 复制代码

mkdir ~/flink && cd ~/flink

wget --no-check-certificate https://archive.apache.org/dist/flink/flink-1.15.3/flink-1.15.3-bin-scala_2.12.tgz 
wget --no-check-certificate https://repo1.maven.org/maven2/com/ververica/flink-sql-connector-mysql-cdc/2.1.1/flink-sql-connector-mysql-cdc-2.1.1.jar
wget --no-check-certificate https://repo.maven.apache.org/maven2/org/apache/flink/flink-connector-jdbc/1.15.3/flink-connector-jdbc-1.15.3.jar
wget --no-check-certificate https://repo.maven.apache.org/maven2/mysql/mysql-connector-java/5.1.47/mysql-connector-java-5.1.47.jar

tar -xf flink-1.15.3-bin-scala_2.12.tgz
cp *.jar flink-1.15.3/lib/
cd flink-1.15.3
chmod -R 777 ./bin/*
./bin/start-cluster.sh
echo http://`hostname -i`:8081/
./bin/sql-client.sh

对 Flink 的理解

Flink 有 cdc 的 connector，有 jdbc 的 connector。其中：

基于日志的CDC：cdc connector 用于做实时同步，jdbc 则用于数据写入。
假设源端使用的是 cdc，那么一个 Flink SQL Job insert into target select from cdc_table 会一直在后台执行，监听数据的变化，并根据变化做计算。
基于查询的CDC：假设源端使用的是 jdbc connector，那么 Flink SQL 会立即执行，读取源端的全部数据并做计算，然后 Job 退出。源端有新的插入，也不会做任何同步操作，因为 Job 已经结束。

// https://cloud.tencent.com/developer/article/2193358

Flink 的市场定位

方案一、Debezium+Kafka+计算程序+存储系统

采用Debezium订阅MySql的Binlog传输到Kafka，后端是由计算程序从kafka里面进行消费，最后将数据写入到其他存储。

方案二、Debezium + Kafka + Flink Sql+存储系统

Flink Sql具备解析 Kafka 中 debezium-json 和 canal-json 格式的 binlog 能力，具体的框架如下

方案三、Flink CDC + JDBC Connector

方案一与方案二的相同点是组件维护复杂，Flink 1.11中CDC Connectors内置了 Debezium 引擎，可以替换 Debeziuum+Kafka.

总结：Flink 的市场定位就是干掉所有传输通道上的人。

通过Flink CDC Connector替换Debezium+Kafka的数据采集模块，实现 Flink Sql 采集+计算+传输（ETL）一体化。优点如下

开箱即用，容易上手
减少维护的组件，简化实时链路，减轻部署成本
减少端到端延迟
Flink 自身支持Exactly Once的读取计算
数据不落地，减少存储成本
支持全量和增量流式读取
binlog采集位点可回溯

最后，Flink 得到下面这样一个架构图：

还有一点薄纱要翻开：Flink SQL 支持什么呢？

sql 复制代码

WITH orders_with_total AS (
    SELECT order_id, price + tax AS total
    FROM Orders
)
SELECT order_id, SUM(total)
FROM orders_with_total
GROUP BY order_id;

SELECT & WHERE

sql 复制代码

SELECT price + tax FROM Orders WHERE id = 10

SELECT DISTINCT

sql 复制代码

SELECT DISTINCT id FROM Orders

Windowing table-valued functions (Windowing TVFs)
Apache Flink provides 3 built-in windowing TVFs: TUMBLE, HOP and CUMULATE. 利用这些 table function，可以把原始表的数据进行分组/扩行。例如下面的例子，用 TUMBLE 把 6 行数据按照 10 分钟的间隔分成了 2 组，然后基于这些组就能做进一步的聚合分析。

Flink 支持等间隔窗口（TUMBLE）、滑动窗口（HOP）、累积窗口（CUMULATE），可以根据实际业务场景选用。比如：

实时统计每个小时的销量，用 TUMBLE 就比较合适，按小时划分间隔。
实时统计最近 60 分钟的销量，则使用 HOP（60min）比较合适，每分钟更新一次。

sql 复制代码

Flink SQL> SELECT * FROM Bid;
+------------------+-------+------+
|          bidtime | price | item |
+------------------+-------+------+
| 2020-04-15 08:05 |  4.00 | C    |
| 2020-04-15 08:07 |  2.00 | A    |
| 2020-04-15 08:09 |  5.00 | D    |
| 2020-04-15 08:11 |  3.00 | B    |
| 2020-04-15 08:13 |  1.00 | E    |
| 2020-04-15 08:17 |  6.00 | F    |
+------------------+-------+------+

Flink SQL> SELECT * FROM TABLE(
   TUMBLE(TABLE Bid, DESCRIPTOR(bidtime), INTERVAL '10' MINUTES));
-- or with the named params
-- note: the DATA param must be the first
Flink SQL> SELECT * FROM TABLE(
   TUMBLE(
     DATA => TABLE Bid,
     TIMECOL => DESCRIPTOR(bidtime),
     SIZE => INTERVAL '10' MINUTES));
+------------------+-------+------+------------------+------------------+-------------------------+
|          bidtime | price | item |     window_start |       window_end |            window_time  |
+------------------+-------+------+------------------+------------------+-------------------------+
| 2020-04-15 08:05 |  4.00 | C    | 2020-04-15 08:00 | 2020-04-15 08:10 | 2020-04-15 08:09:59.999 |
| 2020-04-15 08:07 |  2.00 | A    | 2020-04-15 08:00 | 2020-04-15 08:10 | 2020-04-15 08:09:59.999 |
| 2020-04-15 08:09 |  5.00 | D    | 2020-04-15 08:00 | 2020-04-15 08:10 | 2020-04-15 08:09:59.999 |
| 2020-04-15 08:11 |  3.00 | B    | 2020-04-15 08:10 | 2020-04-15 08:20 | 2020-04-15 08:19:59.999 |
| 2020-04-15 08:13 |  1.00 | E    | 2020-04-15 08:10 | 2020-04-15 08:20 | 2020-04-15 08:19:59.999 |
| 2020-04-15 08:17 |  6.00 | F    | 2020-04-15 08:10 | 2020-04-15 08:20 | 2020-04-15 08:19:59.999 |
+------------------+-------+------+------------------+------------------+-------------------------+

-- apply aggregation on the tumbling windowed table
Flink SQL> SELECT window_start, window_end, SUM(price)
  FROM TABLE(
    TUMBLE(TABLE Bid, DESCRIPTOR(bidtime), INTERVAL '10' MINUTES))
  GROUP BY window_start, window_end;
+------------------+------------------+-------+
|     window_start |       window_end | price |
+------------------+------------------+-------+
| 2020-04-15 08:00 | 2020-04-15 08:10 | 11.00 |
| 2020-04-15 08:10 | 2020-04-15 08:20 | 10.00 |
+------------------+------------------+-------+

Window Aggregate
这就是基于上面的 TVF 的应用。上面 Windowing table-valued functions 提供了数据输入，基于这些数据输入做聚合，就能得到一些统计信息。

sql 复制代码

SELECT ...
FROM <windowed_table> -- relation applied windowing TVF
GROUP BY window_start, window_end, ...


-- hopping window aggregation
Flink SQL> SELECT window_start, window_end, SUM(price)
  FROM TABLE(
    HOP(TABLE Bid, DESCRIPTOR(bidtime), INTERVAL '5' MINUTES, INTERVAL '10' MINUTES))
  GROUP BY window_start, window_end;
+------------------+------------------+-------+
|     window_start |       window_end | price |
+------------------+------------------+-------+
| 2020-04-15 08:00 | 2020-04-15 08:10 | 11.00 |
| 2020-04-15 08:05 | 2020-04-15 08:15 | 15.00 |
| 2020-04-15 08:10 | 2020-04-15 08:20 | 10.00 |
| 2020-04-15 08:15 | 2020-04-15 08:25 | 6.00  |
+------------------+------------------+-------+

GROUPING SET

基于 Window 做一些更高级的窗口函数，按照多种分组方式统计数据。

sql 复制代码

Flink SQL> SELECT window_start, window_end, supplier_id, SUM(price) as price
  FROM TABLE(
    TUMBLE(TABLE Bid, DESCRIPTOR(bidtime), INTERVAL '10' MINUTES))
  GROUP BY window_start, window_end, GROUPING SETS ((supplier_id), ());
+------------------+------------------+-------------+-------+
|     window_start |       window_end | supplier_id | price |
+------------------+------------------+-------------+-------+
| 2020-04-15 08:00 | 2020-04-15 08:10 |      (NULL) | 11.00 |
| 2020-04-15 08:00 | 2020-04-15 08:10 |   supplier2 |  5.00 |
| 2020-04-15 08:00 | 2020-04-15 08:10 |   supplier1 |  6.00 |
| 2020-04-15 08:10 | 2020-04-15 08:20 |      (NULL) | 10.00 |
| 2020-04-15 08:10 | 2020-04-15 08:20 |   supplier2 |  9.00 |
| 2020-04-15 08:10 | 2020-04-15 08:20 |   supplier1 |  1.00 |
+------------------+------------------+-------------+-------+

ROLLUP、CUBE

这些都是 GROUPING SET 的简写，把所有的组合都枚举出来。计算不要钱么？坑爹！ ▄█▀█●
GROUP

这里就是各种分组聚合的语法，包括一般聚合函数，DISTINCT，GROUP BY、GROUPING SET、CUBE、ROLLUP、HAVING 等。

sql 复制代码

SELECT COUNT(*) FROM Orders; //  COUNT, SUM, AVG (average), MAX (maximum) and MIN (minimum) 

SELECT COUNT(DISTINCT order_id) FROM Orders;

SELECT supplier_id, rating, COUNT(*) AS total
FROM (VALUES
    ('supplier1', 'product1', 4),
    ('supplier1', 'product2', 3),
    ('supplier2', 'product3', 3),
    ('supplier2', 'product4', 4))
AS Products(supplier_id, product_id, rating)
GROUP BY GROUPING SETS ((supplier_id, rating), (supplier_id), ());

SELECT SUM(amount)
FROM Orders
GROUP BY users
HAVING SUM(amount) > 50;

OVER () 标准窗口函数

sql 复制代码

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

INNER Equi-JOIN
不是在流上 join，而是和所有数据做 join，过去的、现在的。

sql 复制代码

SELECT * FROM Orders
INNER JOIN Product
ON Orders.productId = Product.id

OUTER Equi-JOIN
不是在流上 join，而是和所有数据做 join，过去的、现在的。

sql 复制代码

SELECT *
FROM Orders
LEFT JOIN Product
ON Orders.product_id = Product.id

SELECT *
FROM Orders
RIGHT JOIN Product
ON Orders.product_id = Product.id

SELECT *
FROM Orders
FULL OUTER JOIN Product
ON Orders.product_id = Product.id

Temporal Joins

sql 复制代码

SELECT 
     order_id,
     price,
     orders.currency,
     conversion_rate,
     order_time
FROM orders
LEFT JOIN currency_rates FOR SYSTEM_TIME AS OF orders.order_time
ON orders.currency = currency_rates.currency;

等等。对于 Flink 来说，Join 是它的重头戏。

Flink 也支持常见的 funciton，字符串的，算术运算的，比较的等等。