Flink SQL Client中的Chained (Event) Time Windows

Flink SQL Cookbook中有Chained (Event) Time Windows

该md文件举了两个例子，

前面一个例子是Chained Windows

后面一个例子是Non-Chained Windows

该md文件的前半部分图示如下:

上图中存在两个server_logs_window_5m

①Chained Windows

②Non-Chained Windows

所以②比①更加有计算上的效率，但是耦合性也更强。

上图图示中，三条红线就是所谓的Chained (Event) Time Windows

这个概念出自Ververcica公司的flink sql cookbook

Chained (Event) Time Windows的SQL距离

Chained (Event) Time Windows的SQL举例(对应上图中的3个红色箭头):

Chained (Event) Time Windows的数据流向	具体对应的SQL
server_logs_window_1m->server_logs_window_5m	CREATE TEMPORARY VIEW server_logs_window_5m AS SELECT TUMBLE_START(window_end, INTERVAL '5' MINUTE) AS window_start, TUMBLE_ROWTIME(window_end, INTERVAL '5' MINUTE) AS window_end, SUM(total_size) AS total_size, SUM(num_requests) AS num_requests FROM server_logs_window_1m GROUP BY TUMBLE(window_end, INTERVAL '5' MINUTE);
server_logs_window_1m->avg_request_size_1m	INSERT INTO avg_request_size_1m SELECT window_start, window_end, total_size/num_requests AS avg_size FROM server_logs_window_1m;
server_logs_window_5m->avg_request_size_5m	INSERT INTO avg_request_size_5m SELECT window_start, window_end, total_size/num_requests AS avg_size FROM server_logs_window_5m;

Chained Windows与Non-Chained Windows的SQL写法对比

Windows类型	具体对应的SQL
Chained Windows	CREATE TEMPORARY VIEW server_logs_window_5m AS SELECT TUMBLE_START(window_end, INTERVAL '5' MINUTE) AS window_start, TUMBLE_ROWTIME(window_end, INTERVAL '5' MINUTE) AS window_end, SUM(total_size) AS total_size, SUM(num_requests) AS num_requests FROM server_logs_window_1m GROUP BY TUMBLE(window_end, INTERVAL '5' MINUTE);
Non-Chained Windows	CREATE TEMPORARY VIEW server_logs_window_5m AS SELECT TUMBLE_START (log_time, INTERVAL '5' MINUTE) AS window_start, TUMBLE_ROWTIME(log_time, INTERVAL '5' MINUTE) AS window_end, SUM(size) AS total_size, COUNT() AS num_requests FROM server_logs* GROUP BY TUMBLE(log_time, INTERVAL '5' MINUTE);

Flink SQL Client中的Chained (Event) Time Windows相关推荐

Flink SQL Client中的session window图解
flink sql cookbook中的sql语句是: SELECT userid,SESSION_START(log_time, INTERVAL '10' SECOND) AS session_b ...
Flink SQL Client注册SCALA UDF完整流程
UDF的完整maven工程與SQL https://github.com/appleyuchi/Flink_SQL_Client_UDF 完整操作步骤 ①mvn scala:compile packa ...
Flink SQL Client方言切换与datagen-＞Hive(DDL形式+streaming形式)
概述本文是对[1]的完整复现,补充了[1]中缺失的大量细节. 切换方言切换目标 Flink SQL Client命令切换为hive SET table.sql-dialect=hive; 切换为 ...
flink sql client讀取kafka數據的timestamp(DDL方式)
实验目的 Kafka的数据能让Flink SQL Client读取到本文是对[1]的详细记载具体操作步骤 ①啓動hadoop集羣,離開安全模式 ②各个节点都关闭防火墙: service firew ...
Flink SQL Client讀取csv中的數據(轉載+總結)
根據官方文檔[2] Flink SQL啓動方式啓動命令 (1)starting an embedded standalone process $FLINK_HOME/bin/sql-client.s ...
Flink SQL Client注册JAVA UDF完整流程
概述听大佬说[1]里面有flink sql client注册udf的方法去看了一眼,全是文字,闹心,索性琢磨了一下,记录下来． UDF的完整maven工程 https://github.com/a ...
Flink SQL Client实现CDC实验
概述本文主要是對[7]中內容的復現环境组件版本 Flink(HA) 1.12 Zookeeper 3.6.0 flink-sql-connector-mysql-cdc 1.1.1 Mysql ...
Flink SQL Client读Kafka+流计算(DDL方式+代碼嵌入DDL/SQL方式)
#################################################################################################### ...
flink sql client读取hive时卡住
问题复现如下: 查看$FLINK_HOME/log/flink-appleyuchi-sql-client-Desktop.log 2020-12-23 11:48:56,811 INFO org. ...

Flink SQL Client中的Chained (Event) Time Windows

Flink SQL Client中的Chained (Event) Time Windows相关推荐

最新文章

热门文章