一、双流JOIN
在Flink中, 支持两种方式的流的Join: Window Join和Interval Join
二、Window Join
窗口join会join具有相同的key并且处于同一个窗口中的两个流的元素.
注意:
1.所有的窗口join都是 inner join, 意味着a流中的元素如果在b流中没有对应的, 则a流中这个元素就不会处理(就是忽略掉了)
2.join成功后的元素的会以所在窗口的最大时间作为其时间戳. 例如窗口[5,10), 则元素会以9作为自己的时间戳。
Window join 仍然可分为 滚动窗口、滑动窗口Join、会话窗口Join
滚动窗口Join代码段示例
sql
package com.lyh.flink12;
import org.apache.flink.api.common.eventtime.SerializableTimestampAssigner;
import org.apache.flink.api.common.eventtime.WatermarkStrategy;
import org.apache.flink.api.common.functions.JoinFunction;
import org.apache.flink.configuration.Configuration;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.windowing.assigners.TumblingEventTimeWindows;
import org.apache.flink.streaming.api.windowing.time.Time;
/**
* @Author lizhenchao@atguigu.cn
* @Date 2021/1/24 22:09
*/
public class Flink01_Join_Window_Tumbling {
public static void main(String[] args) {
StreamExecutionEnvironment env = StreamExecutionEnvironment.createLocalEnvironmentWithWebUI(new Configuration());
env.setParallelism(1);
SingleOutputStreamOperator<WaterSensor> s1 = env
.socketTextStream("hadoop100", 8888) // 在socket终端只输入毫秒级别的时间戳
.map(value -> {
String[] datas = value.split(",");
return new WaterSensor(datas[0], Long.valueOf(datas[1]), Integer.valueOf(datas[2]));
})
.assignTimestampsAndWatermarks(
WatermarkStrategy
.<WaterSensor>forMonotonousTimestamps()
.withTimestampAssigner(new SerializableTimestampAssigner<WaterSensor>() {
@Override
public long extractTimestamp(WaterSensor element, long recordTimestamp) {
return element.getTs() * 1000;
}
})
);
SingleOutputStreamOperator<WaterSensor> s2 = env
.socketTextStream("hadoop100", 9999) // 在socket终端只输入毫秒级别的时间戳
.map(value -> {
String[] datas = value.split(",");
return new WaterSensor(datas[0], Long.valueOf(datas[1]), Integer.valueOf(datas[2]));
})
.assignTimestampsAndWatermarks(
WatermarkStrategy
.<WaterSensor>forMonotonousTimestamps()
.withTimestampAssigner(new SerializableTimestampAssigner<WaterSensor>() {
@Override
public long extractTimestamp(WaterSensor element, long recordTimestamp) {
return element.getTs() * 1000;
}
})
);
s1.join(s2)
.where(WaterSensor::getId)
.equalTo(WaterSensor::getId)
.window(TumblingEventTimeWindows.of(Time.seconds(5))) // 必须使用窗口
.apply(new JoinFunction<WaterSensor, WaterSensor, String>() {
@Override
public String join(WaterSensor first, WaterSensor second) throws Exception {
return "first: " + first + ", second: " + second;
}
})
.print();
try {
env.execute();
} catch (Exception e) {
e.printStackTrace();
}
}
}运行结果：
三、Interval Join
间隔流join(Interval Join), 是指使用一个流的数据按照key去join另外一条流的指定范围的数据.
如下图: 橙色的流去join绿色的流.范围是由橙色流的event-time + lower bound和event-time + upper bound来决定的.
orangeElem.ts + lowerBound <= greenElem.ts <= orangeElem.ts + upperBound
Interval Join只支持event-time
必须是keyBy之后的流才可以interval join
package com.lyh.flink12;
import com.lyh.bean.WaterSensor;
import org.apache.flink.api.common.eventtime.SerializableTimestampAssigner;
import org.apache.flink.api.common.eventtime.WatermarkStrategy;
import org.apache.flink.configuration.Configuration;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.functions.co.ProcessJoinFunction;
import org.apache.flink.streaming.api.windowing.time.Time;
import org.apache.flink.table.planner.expressions.In;
import org.apache.flink.util.Collector;
sql
import java.time.Duration;
public class Sql_Join_Windows_Interval{
public static void main(String[] args) {
StreamExecutionEnvironment env = StreamExecutionEnvironment.createLocalEnvironmentWithWebUI(new Configuration());
env.setParallelism(2);
SingleOutputStreamOperator<WaterSensor> s1 = env.socketTextStream("hadoop100", 8888)
.map(value -> {
String[] data = value.split(",");
return new WaterSensor(
data[0],
Long.valueOf(data[1]),
Integer.valueOf(data[2]));
}).assignTimestampsAndWatermarks(WatermarkStrategy.<WaterSensor>forBoundedOutOfOrderness(Duration.ofSeconds(2))
.withTimestampAssigner(new SerializableTimestampAssigner<WaterSensor>() {
@Override
public long extractTimestamp(WaterSensor element, long timestamp) {
return element.getTs();
}
}));
SingleOutputStreamOperator<WaterSensor> s2 = env.socketTextStream("hadoop100", 9999)
.map(value -> {
String[] data = value.split(",");
return new WaterSensor(
data[0],
Long.valueOf(data[1]),
Integer.valueOf(data[2]));
}).assignTimestampsAndWatermarks(WatermarkStrategy.<WaterSensor>forBoundedOutOfOrderness(Duration.ofSeconds(2))
.withTimestampAssigner(new SerializableTimestampAssigner<WaterSensor>() {
@Override
public long extractTimestamp(WaterSensor element, long timestamp) {
return element.getTs();
}
}));
s1.keyBy(WaterSensor::getId)
.intervalJoin(s2.keyBy(WaterSensor::getId))
.between(Time.seconds(-2),Time.seconds(3))
.process(new ProcessJoinFunction<WaterSensor, WaterSensor, String>() {
@Override
public void processElement(WaterSensor left,
WaterSensor right,
Context ctx,
Collector<String> out) throws Exception {
out.collect(left + "," + right);
}
}).print();
try{
env.execute();
} catch (Exception e)
{e.printStackTrace();}
}
}运行结果：
