58、Spark Streaming: DStream的output操作以及foreachRDD详解

一、output操作

1、output操作

DStream中的所有计算，都是由output操作触发的，比如print()。如果没有任何output操作，那么，压根儿就不会执行定义的计算逻辑。

此外，即使你使用了foreachRDD output操作，也必须在里面对RDD执行action操作，才能触发对每一个batch的计算逻辑。否则，光有
foreachRDD output操作，在里面没有对RDD执行action操作，也不会触发任何逻辑。

2、output操作概览

二、foreachRDD

1、foreachRDD详解

通常在foreachRDD中，都会创建一个Connection，比如JDBC Connection，然后通过Connection将数据写入外部存储。

误区一：在RDD的foreach操作外部，创建Connection

这种方式是错误的，因为它会导致Connection对象被序列化后传输到每个Task中。而这种Connection对象，实际上一般是不支持序列化的，也就无法被传输。

dstream.foreachRDD { rdd =>
  val connection = createNewConnection()
  rdd.foreach { record => connection.send(record)
  }
}

误区二：在RDD的foreach操作内部，创建Connection

这种方式是可以的，但是效率低下。因为它会导致对于RDD中的每一条数据，都创建一个Connection对象。而通常来说，Connection的创建，是很消耗性能的。

dstream.foreachRDD { rdd =>
  rdd.foreach { record =>
    val connection = createNewConnection()
    connection.send(record)
    connection.close()
  }
}

合理方式一：使用RDD的foreachPartition操作，并且在该操作内部，创建Connection对象，这样就相当于是，为RDD的每个partition创建一个Connection对象，节省资源的多了。

dstream.foreachRDD { rdd =>
  rdd.foreachPartition { partitionOfRecords =>
    val connection = createNewConnection()
    partitionOfRecords.foreach(record => connection.send(record))
    connection.close()
  }
}

合理方式二：自己手动封装一个静态连接池，使用RDD的foreachPartition操作，并且在该操作内部，从静态连接池中，通过静态方法，获取到一个连接，
使用之后再还回去。这样的话，甚至在多个RDD的partition之间，也可以复用连接了。而且可以让连接池采取懒创建的策略，并且空闲一段时间后，将其释放掉。

dstream.foreachRDD { rdd =>
  rdd.foreachPartition { partitionOfRecords =>
    val connection = ConnectionPool.getConnection()
    partitionOfRecords.foreach(record => connection.send(record))
    ConnectionPool.returnConnection(connection)
  }
}

案例：改写UpdateStateByKeyWordCount，将每次统计出来的全局的单词计数，写入一份，到MySQL数据库中。

2、java案例

创建mysql表

mysql> use testdb;
Reading table information for completion of table and column names
You can turn off this feature to get a quicker startup with -A

Database changed
mysql> create table wordcount (
    ->   id integer auto_increment primary key,
    ->   updated_time timestamp NOT NULL default CURRENT_TIMESTAMP on update CURRENT_TIMESTAMP,
    ->   word varchar(255),
    ->   count integer
    -> );
Query OK, 0 rows affected (0.05 sec)

java代码

###ConnectionPool 

package cn.spark.study.streaming;

import java.sql.Connection;
import java.sql.DriverManager;
import java.util.LinkedList;

/**
 * 简易版的连接池
 * @author Administrator
 *
 */
public class ConnectionPool {

    // 静态的Connection队列
    private static LinkedList<Connection> connectionQueue;

    /**
     * 加载驱动
     */
    static {
        try {
            Class.forName("com.mysql.jdbc.Driver");
        } catch (ClassNotFoundException e) {
            e.printStackTrace();
        }
    }

    /**
     * 获取连接，多线程访问并发控制
     * @return
     */
    public synchronized static Connection getConnection() {
        try {
            if(connectionQueue == null) {
                connectionQueue = new LinkedList<Connection>();
                for(int i = 0; i < 10; i++) {
                    Connection conn = DriverManager.getConnection(
                            "jdbc:mysql://spark1:3306/testdb",
                            "",
                            "");
                    connectionQueue.push(conn);
                }
            }
        } catch (Exception e) {
            e.printStackTrace();
        }
        return connectionQueue.poll();
    }

    /**
     * 还回去一个连接
     */
    public static void returnConnection(Connection conn) {
        connectionQueue.push(conn);
    }

}

###PersistWordCount

package cn.spark.study.streaming;

import java.sql.Connection;
import java.sql.Statement;
import java.util.Arrays;
import java.util.Iterator;
import java.util.List;

import org.apache.spark.SparkConf;
import org.apache.spark.api.java.JavaPairRDD;
import org.apache.spark.api.java.function.FlatMapFunction;
import org.apache.spark.api.java.function.Function;
import org.apache.spark.api.java.function.Function2;
import org.apache.spark.api.java.function.PairFunction;
import org.apache.spark.api.java.function.VoidFunction;
import org.apache.spark.streaming.Durations;
import org.apache.spark.streaming.api.java.JavaDStream;
import org.apache.spark.streaming.api.java.JavaPairDStream;
import org.apache.spark.streaming.api.java.JavaReceiverInputDStream;
import org.apache.spark.streaming.api.java.JavaStreamingContext;

import com.google.common.base.Optional;

import scala.Tuple2;

/**
 * 基于持久化机制的实时wordcount程序
 * @author Administrator
 *
 */
public class PersistWordCount {

    public static void main(String[] args) {
        SparkConf conf = new SparkConf()
                .setMaster("local[2]")
                .setAppName("PersistWordCount");
        JavaStreamingContext jssc = new JavaStreamingContext(conf, Durations.seconds(5));

        jssc.checkpoint("hdfs://spark1:9000/wordcount_checkpoint");  

        JavaReceiverInputDStream<String> lines = jssc.socketTextStream("spark1", 9999);

        JavaDStream<String> words = lines.flatMap(new FlatMapFunction<String, String>() {

            private static final long serialVersionUID = 1L;

            @Override
            public Iterable<String> call(String line) throws Exception {
                return Arrays.asList(line.split(" "));
            }

        });

        JavaPairDStream<String, Integer> pairs = words.mapToPair(

                new PairFunction<String, String, Integer>() {

                    private static final long serialVersionUID = 1L;

                    @Override
                    public Tuple2<String, Integer> call(String word)
                            throws Exception {
                        return new Tuple2<String, Integer>(word, 1);
                    }

                });

        JavaPairDStream<String, Integer> wordCounts = pairs.updateStateByKey(

                new Function2<List<Integer>, Optional<Integer>, Optional<Integer>>() {

                    private static final long serialVersionUID = 1L;

                    @Override
                    public Optional<Integer> call(List<Integer> values,
                            Optional<Integer> state) throws Exception {
                        Integer newValue = 0;

                        if(state.isPresent()) {
                            newValue = state.get();
                        }

                        for(Integer value : values) {
                            newValue += value;
                        }

                        return Optional.of(newValue);
                    }

                });

        // 每次得到当前所有单词的统计次数之后，将其写入mysql存储，进行持久化，以便于后续的J2EE应用程序
        // 进行显示
        wordCounts.foreachRDD(new Function<JavaPairRDD<String,Integer>, Void>() {

            private static final long serialVersionUID = 1L;

            @Override
            public Void call(JavaPairRDD<String, Integer> wordCountsRDD) throws Exception {
                // 调用RDD的foreachPartition()方法
                wordCountsRDD.foreachPartition(new VoidFunction<Iterator<Tuple2<String,Integer>>>() {

                    private static final long serialVersionUID = 1L;

                    @Override
                    public void call(Iterator<Tuple2<String, Integer>> wordCounts) throws Exception {
                        // 给每个partition，获取一个连接
                        Connection conn = ConnectionPool.getConnection();

                        // 遍历partition中的数据，使用一个连接，插入数据库
                        Tuple2<String, Integer> wordCount = null;
                        while(wordCounts.hasNext()) {
                            wordCount = wordCounts.next();

                            String sql = "insert into wordcount(word,count) "
                                    + "values('" + wordCount._1 + "'," + wordCount._2 + ")";  

                            Statement stmt = conn.createStatement();
                            stmt.executeUpdate(sql);
                        }

                        // 用完以后，将连接还回去
                        ConnectionPool.returnConnection(conn);
                    }
                });

                return null;
            }

        });

        jssc.start();
        jssc.awaitTermination();
        jssc.close();
    }

}

##运行脚本
[root@spark1 streaming]# cat persistWordCount.sh
/usr/local/spark-1.5.1-bin-hadoop2.4/bin/spark-submit \
--class cn.spark.study.streaming.PersistWordCount \
--num-executors 3 \
--driver-memory 100m \
--executor-memory 100m \
--executor-cores 3 \
--files /usr/local/hive/conf/hive-site.xml \
--driver-class-path /usr/local/hive/lib/mysql-connector-java-5.1.17.jar \
/usr/local/spark-study/java/streaming/saprk-study-java-0.0.1-SNAPSHOT-jar-with-dependencies.jar \

##运行nc
[root@spark1 ~]# nc -lk 9999
hello word
hello word
hello java

##结果
mysql> use testdb;
mysql> select * from wordcount;
+----+---------------------+-------+-------+
| id | updated_time        | word  | count |
+----+---------------------+-------+-------+
|  1 | 2019-08-19 14:52:45 | hello |     1 |
|  2 | 2019-08-19 14:52:45 | word  |     1 |
|  3 | 2019-08-19 14:52:50 | hello |     2 |
|  4 | 2019-08-19 14:52:50 | word  |     2 |
|  5 | 2019-08-19 14:52:55 | hello |     2 |
|  6 | 2019-08-19 14:52:55 | word  |     2 |
|  7 | 2019-08-19 14:53:00 | hello |     2 |
|  8 | 2019-08-19 14:53:00 | word  |     2 |
|  9 | 2019-08-19 14:53:05 | hello |     2 |
| 10 | 2019-08-19 14:53:05 | word  |     2 |
| 11 | 2019-08-19 14:53:10 | hello |     2 |
| 12 | 2019-08-19 14:53:10 | word  |     2 |
| 13 | 2019-08-19 14:53:15 | hello |     3 |
| 14 | 2019-08-19 14:53:15 | word  |     2 |
| 15 | 2019-08-19 14:53:15 | java  |     1 |
| 16 | 2019-08-19 14:53:20 | hello |     3 |
| 17 | 2019-08-19 14:53:20 | word  |     2 |
| 18 | 2019-08-19 14:53:20 | java  |     1 |
| 19 | 2019-08-19 14:53:25 | hello |     3 |
| 20 | 2019-08-19 14:53:25 | word  |     2 |
| 21 | 2019-08-19 14:53:25 | java  |     1 |
+----+---------------------+-------+-------+
21 rows in set (0.00 sec)