Kafka Java客戶端代碼的示例分析

這篇文章將為大家詳細(xì)講解有關(guān)Kafka Java客戶端代碼的示例分析，文章內(nèi)容質(zhì)量較高，因此小編分享給大家做個(gè)參考，希望大家閱讀完這篇文章后對(duì)相關(guān)知識(shí)有一定的了解。

kafka是一種高吞吐量的分布式發(fā)布訂閱消息系統(tǒng)

kafka是linkedin用于日志處理的分布式消息隊(duì)列，linkedin的日志數(shù)據(jù)容量大，但對(duì)可靠性要求不高，其日志數(shù)據(jù)主要包括用戶行為（登錄、瀏覽、點(diǎn)擊、分享、喜歡）以及系統(tǒng)運(yùn)行日志（CPU、內(nèi)存、磁盤、網(wǎng)絡(luò)、系統(tǒng)及進(jìn)程狀態(tài)）

當(dāng)前很多的消息隊(duì)列服務(wù)提供可靠交付保證，并默認(rèn)是即時(shí)消費(fèi)（不適合離線）。

高可靠交付對(duì)linkedin的日志不是必須的，故可通過降低可靠性來(lái)提高性能，同時(shí)通過構(gòu)建分布式的集群，允許消息在系統(tǒng)中累積，使得kafka同時(shí)支持離線和在線日志處理

測(cè)試環(huán)境

kafka_2.10-0.8.1.1 3個(gè)節(jié)點(diǎn)做的集群

zookeeper-3.4.5 一個(gè)實(shí)例節(jié)點(diǎn)

代碼示例

消息生產(chǎn)者代碼示例

import java.util.Collections;  import java.util.Date;  import java.util.Properties;  import java.util.Random;     import kafka.javaapi.producer.Producer;  import kafka.producer.KeyedMessage;  import kafka.producer.ProducerConfig;     /**   * 詳細(xì)可以參考：https://cwiki.apache.org/confluence/display/KAFKA/0.8.0+Producer+Example   * @author Fung   *   */ public class ProducerDemo {      public static void main(String[] args) {          Random rnd = new Random();          int events=100;             // 設(shè)置配置屬性          Properties props = new Properties();          props.put("metadata.broker.list","172.168.63.221:9092,172.168.63.233:9092,172.168.63.234:9092");          props.put("serializer.class", "kafka.serializer.StringEncoder");          // key.serializer.class默認(rèn)為serializer.class          props.put("key.serializer.class", "kafka.serializer.StringEncoder");          // 可選配置，如果不配置，則使用默認(rèn)的partitioner          props.put("partitioner.class", "com.catt.kafka.demo.PartitionerDemo");          // 觸發(fā)acknowledgement機(jī)制，否則是fire and forget，可能會(huì)引起數(shù)據(jù)丟失          // 值為0,1,-1,可以參考          // http://kafka.apache.org/08/configuration.html          props.put("request.required.acks", "1");          ProducerConfig config = new ProducerConfig(props);             // 創(chuàng)建producer          Producer<String, String> producer = new Producer<String, String>(config);          // 產(chǎn)生并發(fā)送消息          long start=System.currentTimeMillis();          for (long i = 0; i < events; i++) {              long runtime = new Date().getTime();              String ip = "192.168.2." + i;//rnd.nextInt(255);              String msg = runtime + ",www.example.com," + ip;              //如果topic不存在，則會(huì)自動(dòng)創(chuàng)建，默認(rèn)replication-factor為1，partitions為0              KeyedMessage<String, String> data = new KeyedMessage<String, String>(                      "page_visits", ip, msg);              producer.send(data);          }          System.out.println("耗時(shí):" + (System.currentTimeMillis() - start));          // 關(guān)閉producer          producer.close();      }  }

消息消費(fèi)者代碼示例

import java.util.HashMap;  import java.util.List;  import java.util.Map;  import java.util.Properties;  import java.util.concurrent.ExecutorService;  import java.util.concurrent.Executors;     import kafka.consumer.Consumer;  import kafka.consumer.ConsumerConfig;  import kafka.consumer.KafkaStream;  import kafka.javaapi.consumer.ConsumerConnector;     /**   * 詳細(xì)可以參考：https://cwiki.apache.org/confluence/display/KAFKA/Consumer+Group+Example   *    * @author Fung   *   */ public class ConsumerDemo {      private final ConsumerConnector consumer;      private final String topic;      private ExecutorService executor;         public ConsumerDemo(String a_zookeeper, String a_groupId, String a_topic) {          consumer = Consumer.createJavaConsumerConnector(createConsumerConfig(a_zookeeper,a_groupId));          this.topic = a_topic;      }         public void shutdown() {          if (consumer != null)              consumer.shutdown();          if (executor != null)              executor.shutdown();      }         public void run(int numThreads) {          Map<String, Integer> topicCountMap = new HashMap<String, Integer>();          topicCountMap.put(topic, new Integer(numThreads));          Map<String, List<KafkaStream<byte[], byte[]>>> consumerMap = consumer                  .createMessageStreams(topicCountMap);          List<KafkaStream<byte[], byte[]>> streams = consumerMap.get(topic);             // now launch all the threads          executor = Executors.newFixedThreadPool(numThreads);             // now create an object to consume the messages          //          int threadNumber = 0;          for (final KafkaStream stream : streams) {              executor.submit(new ConsumerMsgTask(stream, threadNumber));              threadNumber++;          }      }         private static ConsumerConfig createConsumerConfig(String a_zookeeper,              String a_groupId) {          Properties props = new Properties();          props.put("zookeeper.connect", a_zookeeper);          props.put("group.id", a_groupId);          props.put("zookeeper.session.timeout.ms", "400");          props.put("zookeeper.sync.time.ms", "200");          props.put("auto.commit.interval.ms", "1000");             return new ConsumerConfig(props);      }         public static void main(String[] arg) {          String[] args = { "172.168.63.221:2188", "group-1", "page_visits", "12" };          String zooKeeper = args[0];          String groupId = args[1];          String topic = args[2];          int threads = Integer.parseInt(args[3]);             ConsumerDemo demo = new ConsumerDemo(zooKeeper, groupId, topic);          demo.run(threads);             try {              Thread.sleep(10000);          } catch (InterruptedException ie) {             }          demo.shutdown();      }  }

消息處理類

import kafka.consumer.ConsumerIterator;  import kafka.consumer.KafkaStream;     public class ConsumerMsgTask implements Runnable {      private KafkaStream m_stream;      private int m_threadNumber;         public ConsumerMsgTask(KafkaStream stream, int threadNumber) {          m_threadNumber = threadNumber;          m_stream = stream;      }         public void run() {          ConsumerIterator<byte[], byte[]> it = m_stream.iterator();          while (it.hasNext())              System.out.println("Thread " + m_threadNumber + ": "                     + new String(it.next().message()));          System.out.println("Shutting down Thread: " + m_threadNumber);      }  }

Partitioner類示例

import kafka.producer.Partitioner;  import kafka.utils.VerifiableProperties;     public class PartitionerDemo implements Partitioner {      public PartitionerDemo(VerifiableProperties props) {         }         @Override     public int partition(Object obj, int numPartitions) {          int partition = 0;          if (obj instanceof String) {              String key=(String)obj;              int offset = key.lastIndexOf('.');              if (offset > 0) {                  partition = Integer.parseInt(key.substring(offset + 1)) % numPartitions;              }          }else{              partition = obj.toString().length() % numPartitions;          }                     return partition;      }     }

關(guān)于Kafka Java客戶端代碼的示例分析就分享到這里了，希望以上內(nèi)容可以對(duì)大家有一定的幫助，可以學(xué)到更多知識(shí)。如果覺得文章不錯(cuò)，可以把它分享出去讓更多的人看到。