kafka与streaming集成两种方式

hadoop,spark,kafka交流群：224209501

标签（空格分隔）： spark

简介

Apache Kafka是分布式发布-订阅消息系统。它最初由Linkedln公司开发，之后成为Apache项目的一部分。Kafka是一种快速、可扩展、设计内在就是分布式的，分区的和可复制的提交日志服务。

Apache Kafka与传统消息系统相比有以下不同：

它被设计为一个分布式系统，易于向外扩展

他同时为发布和订阅提供高吞吐量

它支持多订阅者，当失败时能自动平衡消费者；

它将消息持久化到磁盘，因此可用于批量消费，例如ETL，以及实时应用程序。

kafka架构

生产者(Producer)是能够发布消息到话题的任何对象。

已发布的消息保存在一组服务器中，他们被称为代理(Broker)或者Kafka集群

消费者可以订阅一个或多个话题，并从Broker拉数据，从而消费这些已发布的消息。

话题(Topic)是特定类型的消息流。消息是字节的有效负载Payload)，话题是消息的分类或者种子(feed)名。

kafka集群模式单集群



详细的kafka结构图



A two server Kafka cluster hosting four partitions (P0-P3) with two consumer groups. Consumer group A has two consumer instances and group B has four.

1，配置kafka

1，produce.properties

metadata.broker.list=miaodonghua.host:9092

2，server.properties

host.name=miaodonghua.host
log.dirs=/opt/cdh5.3.6/kafka_2.10-0.8.2.1/kafka-logs
zookeeper.connect=miaodonghua.host:2181

3，使用kafka

bin/kafka-server-start.sh config/server.properties

1) 创建 Topic

bin/kafka-topics.sh --create --zookeeper miaodonghua.host:2181 --replication-factor 1 --partitions 1 --topic test

查看topic

bin/kafka-topics.sh --list --zookeeper miaodonghua.host:2181

2) 发布信息到Topic

bin/kafka-console-producer.sh --broker-list miaodonghua.host:9092 --topic ucloudSafe

3) 订阅者订阅消息

bin/kafka-console-consumer.sh --zookeeper localhost:2181 --topic test --from-beginning

2，Receiver-based Approach

1，启动spark-shell

bin/spark-shell \
--jars /opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externallibs/spark-streaming-kafka_2.10-1.5.2.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externallibs/kafka_2.10-0.8.2.2.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externallibs/kafka-clients-0.8.2.2.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externallibs/metrics-core-2.2.0.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externallibs/zkclient-0.3.jar \
--master local[2]

2，编写kafkaWordCount.scala

Approach 1: Receiver-based Approach

import org.apache.spark._
import org.apache.spark.streaming._
import org.apache.spark.streaming.kafka._

val ssc = new StreamingContext(sc, Seconds(5))

val topicMap = Map("test" -> 1)

val lines = KafkaUtils.createStream(ssc, "miaodonghua.host:2181", "testWordCountGroup", topicMap).map(_._2)

val words = lines.flatMap(_.split(" "))
val counts = words.map((_, 1L)).reduceByKey(_ + _)

counts.print()

ssc.start()
ssc.awaitTermination()

3，执行脚本

:load /opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/kakaWordCount.scala

3，Direct Approach

1，启动spark-shell

bin/spark-shell \
--jars /opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externaljars/spark-streaming-kafka_2.10-1.3.0.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externaljars/kafka_2.10-0.8.2.1.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externaljars/kafka-clients-0.8.2.1.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externaljars/metrics-core-2.2.0.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externaljars/zkclient-0.3.jar \
--master local[2]

2，编写kafkaWordCount2.scala

Approach 2: (No Receivers)

import kafka.serializer.StringDecoder
import org.apache.spark._
import org.apache.spark.streaming._
import org.apache.spark.streaming.kafka._

val ssc = new StreamingContext(sc, Seconds(5))

val kafkaMapParams = Map("metadata.broker.list" -> "miaodonghua.host:9092")
val topicsSet = Set("test")

val lines = KafkaUtils.createDirectStream[String, String, StringDecoder, StringDecoder](ssc, kafkaMapParams, topicsSet).map(_._2)

val words = lines.flatMap(_.split(" "))
val counts = words.map((_, 1L)).reduceByKey(_ + _)

counts.print()

ssc.start()
ssc.awaitTermination()

3，执行脚本

:load /opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/kakaWordCount2.scala

![kafka接受数据2.png-29.3kB][9]

4，UpdateStateByKey

1，启动spark-shell

bin/spark-shell \
--jars /opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externaljars/spark-streaming-kafka_2.10-1.3.0.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externaljars/kafka_2.10-0.8.2.1.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externaljars/kafka-clients-0.8.2.1.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externaljars/metrics-core-2.2.0.jar,\
/opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/externaljars/zkclient-0.3.jar \
--master local[2]

2，UpdateStateByKey

import kafka.serializer.StringDecoder
import org.apache.spark._
import org.apache.spark.streaming._
import org.apache.spark.streaming.kafka._

val ssc = new StreamingContext(sc, Seconds(5))
ssc.checkpoint(".")

val kafkaMapParams = Map("metadata.broker.list" -> "miaodonghua.host:9092")
val topicsSet = Set("test")

// Option[S]
val updateFunc =(values: Seq[Int], state: Option[Int]) => {
val currentCount = values.sum
val previousCount = state.getOrElse(0)
Some(currentCount + previousCount)
}

val lines = KafkaUtils.createDirectStream[String, String, StringDecoder, StringDecoder](ssc, kafkaMapParams, topicsSet).map(_._2)

val words = lines.flatMap(_.split(" "))
val counts = words.map((_, 1)) // (hello,list(1,2,3,4,5)) (spark,1)

// updateStateByKey
val state = counts.updateStateByKey[Int](updateFunc)

state.print()

ssc.start()
ssc.awaitTermination()

3，执行脚本

:load /opt/cdh5.3.6/spark-1.3.0-bin-2.5.0-cdh5.3.6/UpdateStateByKey.scala