1flink理论批处理与流处理+简单示例

【README】

1.本文包含了 批处理与流处理的代码示例&＃xff1b;

• 批处理&＃xff1a;把数据 攒在一起&＃xff08;或攒一段时间或攒一定内存大小&＃xff09;&＃xff0c;然后再处理&＃xff0c;这叫批处理&＃xff1b;
• 流处理&＃xff1a;数据每来一个就处理一个&＃xff1b;

2.特点&＃xff1a;

3.引入flink的maven依赖&＃xff1a;

org.apache.flinkflink-java1.14.4org.apache.flinkflink-streaming-java_2.121.14.4org.apache.flinkflink-clients_2.121.14.4

【1】flink批处理离线数据&＃xff08;数据有限&＃xff09;

【1.1】代码

1&＃xff09;数据源&＃xff0c;我们保存在本地文本文件中&＃xff0c;命名为  hello.txt

hello world
hello flink
how are you
thank you
hello zhangsan
hello lisi

2&＃xff09;批处理代码&＃xff1a;

/*** &＃64;Description 批处理&＃xff0c;word count程序&＃xff08;离线数据&＃xff09;* &＃64;author xiao tang* &＃64;version 1.0.0* &＃64;createTime 2022年04月09日*/
public class WordCount {public static void main(String[] args) throws Exception {// 创建执行环境ExecutionEnvironment env &＃61; ExecutionEnvironment.getExecutionEnvironment();// 从文件中读取数据String inputPath &＃61; "D:\\workbench_idea\\diydata\\flinkdemo2\\src\\main\\resources\\hello.txt";DataSource dataSource &＃61; env.readTextFile(inputPath);// 对数据集处理&＃xff0c;按照空格分词展开&＃xff0c;转为 (word,1) 二元组统计DataSet> resultSet &＃61; dataSource.flatMap(new MyFlatMapper()).groupBy(0) // 按照第1个位置的word分组.sum(1); // 将第2个位置上的数据求和resultSet.print();}public static class MyFlatMapper implements FlatMapFunction> {&＃64;Overridepublic void flatMap(String value, Collector> collector) throws Exception {// 按照空格分词String[] words &＃61; value.split(" ");// 遍历所有word&＃xff0c;包装成word 输出Arrays.stream(words).forEach(x->{collector.collect(new Tuple2<>(x, 1));});}}
}

批处理打印结果&＃xff1a;

(you,2)
(flink,1)
(world,1)
(hello,4)
(lisi,1)
(zhangsan,1)
(are,1)
(thank,1)
(how,1)

批处理的结果是最终结果&＃xff1b;

【2】flink流处理离线数据&＃xff08;数据有限&＃xff09;

/*** &＃64;Description 流数据&＃xff08;无限数据&＃xff09;* &＃64;author xiao tang* &＃64;version 1.0.0* &＃64;createTime 2022年04月09日*/
public class StreamWordCount {public static void main(String[] args) throws Exception {// 流处理执行环境StreamExecutionEnvironment streamEnv &＃61; StreamExecutionEnvironment.getExecutionEnvironment();streamEnv.setParallelism(2); // 设置并行度// 从文件中读取数据String inputPath &＃61; "D:\\workbench_idea\\diydata\\flinkdemo2\\src\\main\\resources\\hello.txt";DataStream dataStream &＃61; streamEnv.readTextFile(inputPath);// 定义流操作DataStream> resultStream &＃61; dataStream.flatMap(new WordCount.MyFlatMapper()).keyBy(0).sum(1);// 打印结果resultStream.print();// 执行任务&＃xff08;流终止操作&＃xff09;streamEnv.execute();}
}

打印结果&＃xff1a;

2> (world,1)
1> (thank,1)
2> (flink,1)
1> (hello,1)
2> (how,1)
2> (you,1)
1> (hello,2)
2> (you,2)
1> (hello,3)
2> (zhangsan,1)
1> (hello,4)
2> (lisi,1)
1> (are,1)

流处理的结果是一个动态变化的有状态的结果&＃xff1b;

有状态的意思说白了就是&＃xff1a;后面的处理结果依赖前面的处理结果&＃xff0c;如对hello计数为3&＃xff0c;它是在前面hello计数为2的基础上做的处理&＃xff1b;

【3】flink流处理在线数据&＃xff08;数据无限&＃xff09;

我们引入了 netcat&＃xff08;nc&＃xff09;&＃xff0c;底层使用socket模拟向某端口写入数据&＃xff1b;

然后 flink监控该端口的数据&＃xff0c;并做处理&＃xff1b;

【3.1】 flink处理类

处理类监听了 nc所在机器的的端口&＃xff0c;即 192.168.163.201:7777&＃xff1b;

/*** &＃64;Description socket文本流词计数* &＃64;author xiao tang* &＃64;version 1.0.0* &＃64;createTime 2022年04月09日*/
public class SocketTextStreamWordCount {public static void main(String[] args) throws Exception {// 流处理执行环境StreamExecutionEnvironment streamEnv &＃61; StreamExecutionEnvironment.getExecutionEnvironment();streamEnv.setParallelism(2); // 设置并行度// 从 flinkjava parametertool 获取参数&＃xff08;或有&＃xff09;
// ParameterTool parameterTool &＃61; ParameterTool.fromArgs(args);
// String host &＃61; parameterTool.get("host");
// int port &＃61; parameterTool.getInt("port");// 从socket文本流读取数据DataStream inputDataStream &＃61; streamEnv.socketTextStream("192.168.163.201", 7777);// 定义流操作DataStream> resultStream &＃61; inputDataStream.flatMap(new WordCount.MyFlatMapper()).keyBy(0).sum(1);// 打印结果resultStream.print();// 执行任务&＃xff08;流终止操作&＃xff09;streamEnv.execute();}
}

演示效果&＃xff1a;