Flink sql 窗口函数功能验证

• 数据准备:向测试环境kafka TESTtopic依次添加如下6条数据

  {
     "bidtime":"2020-04-15 08:05:00",
     "price":"4.00",
     "item":"C"
  },
  {
      "bidtime":"2020-04-15 08:07:00",
      "price":"2.00",
      "item":"A"
   },
   {
      "bidtime":"2020-04-15 08:09:00",
      "price":"5.00",
      "item":"D"
   },
   {
      "bidtime":"2020-04-15 08:11:00",
      "price":"3.00",
      "item":"B"
   },
   {
      "bidtime":"2020-04-15 08:13:00",
      "price":"1.00",
      "item":"E"
   },
   {
      "bidtime":"2020-04-15 08:17:00",
      "price":"6.00",
      "item":"F"
   }
  

• 打开flink sql client 建表:

  • 事件时间(Event Time)

   create table Bid(
  	bidtime TIMESTAMP(3),
  	price   DECIMAL(10,2),
  	item    STRING,
  	watermark for bidtime as bidtime - interval '1' second
  )
  WITH ('connector' = 'kafka', 'topic' ='TEST.LIQIANG',
         'properties.bootstrap.servers' ='172.22.17.26:9092,172.22.17.27:9092,172.22.17.28:9092',
  	   'scan.startup.mode' = 'earliest-offset',
         'properties.group.id'='flink-sql-client-local',
         'properties.fetch.max.bytes'='5242880', 
  	   'format' = 'json');
  

  • 处理时间(Processing Time)

   create table Bid(
  	bidtime TIMESTAMP(3),
  	price   DECIMAL(10,2),
  	item    STRING,
  	pt as PROCTIME()
  )
  WITH ('connector' = 'kafka', 'topic' ='TEST.LIQIANG',
         'properties.bootstrap.servers' ='172.22.17.26:9092,172.22.17.27:9092,172.22.17.28:9092',
  	   'scan.startup.mode' = 'earliest-offset',
         'properties.group.id'='flink-sql-client-local',
         'properties.fetch.max.bytes'='5242880', 
  	   'format' = 'json');
  

• 做窗口查询:

  • 事件时间查询

   select tumble_rowtime(bidtime,interval '20' SECOND) as window_time,sum(price)
   from Bid
   group by tumble(bidtime,interval '20' SECOND);
  

  • 处理时间查询

   select 
   TUMBLE_PROCTIME(pt,interval '20' SECOND) as window_time,
   TUMBLE_START(pt,interval '20' SECOND) as window_start,
   TUMBLE_START(pt,interval '20' SECOND) as window_end,
   sum(price) as sum_price
   from Bid
   group by tumble(pt,interval '20' SECOND);
  

• 结果展示: