自学内容网 自学内容网

Flink维表join

所谓的维表Join: 进入Flink的数据,需要关联另外一些存储设备的数据,才能计算出来结果,那么存储在外部设备上的表称之为维表,可能存储在mysql也可能存储在hbase 等。维表一般的特点是变化比较慢。

需求:kafka输入的数据格式: 姓名,城市编号 例如 zhangsan,1001。

期望输出的数据: 姓名,城市编号,城市名称 例如 zhangsan,1001,北京

在MySQL创建城市表作为维表:

create table city(
  city_id varchar(50) primary key,
  city_name varchar(50) 
);
insert into city values('1001','北京'),('1002','上海'),('1003','郑州') ;

1、 预加载维表

通过定义一个类实现RichMapFunction,在open()中读取维表数据加载到内存中,在kafka流map()方法中与维表数据进行关联。

RichMapFunction中open方法里加载维表数据到内存的方式特点如下:

  • 优点:实现简单
  • 缺点:因为数据存于内存,所以只适合小数据量并且维表数据更新频率不高的情况下。虽然可以在open中定义一个定时器定时更新维表,但是还是存在维表更新不及时的情况。另外,维表是变化慢,不是一直不变的,只是变化比较缓慢而已。
package com.bigdata.day06;

import org.apache.flink.api.common.RuntimeExecutionMode;
import org.apache.flink.api.common.functions.RichMapFunction;
import org.apache.flink.api.java.tuple.Tuple3;
import org.apache.flink.configuration.Configuration;
import org.apache.flink.streaming.api.datastream.DataStreamSource;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;

import java.sql.Connection;
import java.sql.DriverManager;
import java.sql.PreparedStatement;
import java.sql.ResultSet;
import java.util.HashMap;
import java.util.Map;


public class _04PreLoadDataDemo {

    public static void main(String[] args) throws Exception {

        //1. env-准备环境
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.setRuntimeMode(RuntimeExecutionMode.AUTOMATIC);

        //2. source-加载数据
        DataStreamSource<String> dataStreamSource = env.socketTextStream("localhost", 9999);
        //3. transformation-数据处理转换
        dataStreamSource.map(new RichMapFunction<String, Tuple3<String,Integer,String>>() {

            Map<Integer,String> cityMap = new HashMap<Integer,String>();
            Connection connection;
            PreparedStatement statement;
            @Override
            public void open(Configuration parameters) throws Exception {
                // 将mysql的数据加载到map中
                connection = DriverManager.getConnection("jdbc:mysql://localhost:3306/test1","root","123456");
                statement = connection.prepareStatement("select * from city");

                ResultSet resultSet = statement.executeQuery();
                while(resultSet.next()){
                   int cityId =  resultSet.getInt("city_id");
                   String cityName =  resultSet.getString("city_name");
                   cityMap.put(cityId,cityName);
                }
            }

            @Override
            public void close() throws Exception {
                statement.close();
                connection.close();
            }

            // zhangsan,1001
            @Override
            public Tuple3<String, Integer, String> map(String s) throws Exception {


                String[] arr = s.split(",");
                System.out.println("+++++++++++++++" +cityMap);
                String cityName = cityMap.get(Integer.valueOf(arr[1]));

                return Tuple3.of(arr[0],Integer.valueOf(arr[1]),cityName);
            }
        }).print();
        //4. sink-数据输出


        //5. execute-执行
        env.execute();
    }
}

测试:

在黑窗口输入:
张三,1001
李四,1001
王五,1002

 那如果数据多了怎么办,数据更新了怎么办?可以进行查询,代码示例如下:

package com.bigdata.day06;

import org.apache.flink.api.common.RuntimeExecutionMode;
import org.apache.flink.api.common.functions.RichMapFunction;
import org.apache.flink.api.java.tuple.Tuple3;
import org.apache.flink.configuration.Configuration;
import org.apache.flink.streaming.api.datastream.DataStreamSource;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;

import java.sql.Connection;
import java.sql.DriverManager;
import java.sql.PreparedStatement;
import java.sql.ResultSet;
import java.util.HashMap;
import java.util.Map;


public class _05SelectDBDemo {

    public static void main(String[] args) throws Exception {

        //1. env-准备环境
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.setRuntimeMode(RuntimeExecutionMode.AUTOMATIC);

        //2. source-加载数据
        DataStreamSource<String> dataStreamSource = env.socketTextStream("localhost", 9999);
        //3. transformation-数据处理转换
        dataStreamSource.map(new RichMapFunction<String, Tuple3<String,Integer,String>>() {


            Connection connection;
            PreparedStatement statement;
            @Override
            public void open(Configuration parameters) throws Exception {
                // 将mysql的数据加载到map中
                connection = DriverManager.getConnection("jdbc:mysql://localhost:3306/test1","root","123456");
                statement = connection.prepareStatement("select city_name from city where city_id = ? ");


            }

            @Override
            public void close() throws Exception {
                statement.close();
                connection.close();
            }

            // zhangsan,1001
            @Override
            public Tuple3<String, Integer, String> map(String s) throws Exception {


                String[] arr = s.split(",");

                statement.setInt(1,Integer.valueOf(arr[1]));
                ResultSet resultSet = statement.executeQuery();
                String cityName = null;
                if(resultSet.next()){
                    cityName = resultSet.getString("city_name");
                }
                return Tuple3.of(arr[0],Integer.valueOf(arr[1]),cityName);
            }
        }).print();
        //4. sink-数据输出


        //5. execute-执行
        env.execute();
    }
}

以上做法成功解决了我们以前的两个问题:数据更新怎么办,数据多了怎么办。

但是缺点是每次都得查询数据库,非常不方便。

2、 热存储维表 

以前的方式是将维表数据存储在Redis、HBase、MySQL等外部存储中,实时流在关联维表数据的时候实时去外部存储中查询,这种方式特点如下:

  • 优点:维度数据量不受内存限制,可以存储很大的数据量。
  • 缺点:因为维表数据在外部存储中,读取速度受制于外部存储的读取速度;另外维表的同步也有延迟。

(1) 使用cache来减轻访问压力

可以使用缓存来存储一部分常访问的维表数据,以减少访问外部系统的次数,比如使用Guava Cache。

package com.bigdata.day06;

import org.apache.flink.api.common.RuntimeExecutionMode;
import org.apache.flink.api.common.functions.RichMapFunction;
import org.apache.flink.api.java.tuple.Tuple3;
import org.apache.flink.configuration.Configuration;
import org.apache.flink.shaded.guava18.com.google.common.cache.*;
import org.apache.flink.streaming.api.datastream.DataStreamSource;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;

import java.sql.Connection;
import java.sql.DriverManager;
import java.sql.PreparedStatement;
import java.sql.ResultSet;
import java.util.concurrent.TimeUnit;


public class _06GuavaCacheDemo {

    public static void main(String[] args) throws Exception {

        //1. env-准备环境
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.setRuntimeMode(RuntimeExecutionMode.AUTOMATIC);
        // 将程序的并行度设置为1,能够更好的展示缓存效果
        env.setParallelism(1);

        //2. source-加载数据
        DataStreamSource<String> dataStreamSource = env.socketTextStream("localhost", 9999);
        //3. transformation-数据处理转换
        dataStreamSource.map(new RichMapFunction<String, Tuple3<String,Integer,String>>() {


            Connection connection;
            PreparedStatement statement;

            // 定义一个Cache
            LoadingCache<Integer, String> cache;
            @Override
            public void open(Configuration parameters) throws Exception {

                cache = CacheBuilder.newBuilder()
                        //最多缓存个数,超过了就根据最近最少使用算法来移除缓存 LRU
                        .maximumSize(1000)
                        //在更新后的指定时间后就回收
                        // 不会自动调用,而是当过期后,又用到了过期的key值数据才会触发的。
                        .expireAfterWrite(10, TimeUnit.SECONDS)
                        //指定移除通知
                        .removalListener(new RemovalListener<Integer, String>() {
                            @Override
                            public void onRemoval(RemovalNotification<Integer, String> removalNotification) {
                                System.out.println(removalNotification.getKey() + "被移除了,值为:" + removalNotification.getValue());
                            }
                        })
                        .build(//指定加载缓存的逻辑
                                new CacheLoader<Integer, String>() {
                                    // 假如缓存中没有数据,会触发该方法的执行,并将结果自动保存到缓存中
                                    @Override
                                    public String load(Integer cityId) throws Exception {
                                        System.out.println("进入数据库查询啦。。。。。。。");
                                        statement.setInt(1,cityId);
                                        ResultSet resultSet = statement.executeQuery();
                                        String cityName = null;
                                        if(resultSet.next()){
                                            System.out.println("进入到了if中.....");
                                            cityName = resultSet.getString("city_name");
                                        }
                                        return cityName;
                                    }
                                });
                // 将mysql的数据加载到map中
                connection = DriverManager.getConnection("jdbc:mysql://localhost:3306/test1","root","123456");
                statement = connection.prepareStatement("select city_name from city where city_id = ? ");
            }

            @Override
            public void close() throws Exception {
                statement.close();
                connection.close();
            }

            // zhangsan,1001
            @Override
            public Tuple3<String, Integer, String> map(String s) throws Exception {

                String[] arr = s.split(",");
                String cityName = "" ;
                if (cache.get(Integer.valueOf(arr[1])) != null){
                    cityName = cache.get(Integer.valueOf(arr[1]));
                }

                return Tuple3.of(arr[0],Integer.valueOf(arr[1]),cityName);
            }
        }).print();
        //4. sink-数据输出


        //5. execute-执行
        env.execute();
    }
}

 设置的guawa缓存是每一个分区都有一个缓存,多个分区之间缓存不共享。所以你需要把并行度设置为1,方便查看效果。

 

 


原文地址:https://blog.csdn.net/weixin_63297999/article/details/144176120

免责声明:本站文章内容转载自网络资源,如本站内容侵犯了原著者的合法权益,可联系本站删除。更多内容请关注自学内容网(zxcms.com)!