(一)理论基础
更多理论以后再补充,或者参考书籍
1、trident是什么?
Trident is a high-level abstraction for doing realtime computing on top of Storm. It allows you to seamlessly intermix high throughput (millions of messages per second), stateful stream processing with low latency distributed querying. If you're familiar with high level batch processing tools like Pig or Cascading, the concepts of Trident will be very familiar – Trident has joins, aggregations, grouping, functions, and filters. In addition to these, Trident adds primitives for doing stateful, incremental processing on top of any database or persistence store. Trident has consistent, exactly-once semantics, so it is easy to reason about Trident topologies.
简单的说,trident是storm的更高层次抽象,相对storm,它主要提供了2个方面的好处:
(1)提供了更高层次的抽象,将常用的count,sum等封装成了方法,可以直接调用,不需要自己实现。
(2)提供了一次原语,如groupby等。
(3)提供了事务支持,可以保证数据均处理且只处理了一次。
2、trident每次处理消息均为batch为单位,即一次处理多个元组。
3、事务类型
关于事务类型,有2个比较容易混淆的概念:spout的事务类型以及事务状态。
它们都有3种类型,分别为:事务型、非事务型和透明事务型。
(1)spout
spout的类型指定了由于下游出现问题导致元组需要重放时,应该怎么发送元组。
事务型spout:重放时能保证同一个批次发送同一批元组。可以保证每一个元组都被发送且只发送一个,且同一个批次所发送的元组是一样的。
非事务型spout:没有任何保障,发完就算。
透明事务型spout:同一个批次发送的元组有可能不同的,它可以保证每一个元组都被发送且只发送一次,但不能保证重放时同一个批次的数据是一样的。这对于部分失效的情况尤其有用,假如以kafka作为spout,当一个topic的某个分区失效时,可以用其它分区的数据先形成一个批次发送出去,如果是事务型spout,则必须等待那个分区恢复后才能继续发送。
这三种类型可以分别通过实现ITransactionalSpout、ITridentSpout、IOpaquePartitionedTridentSpout接口来定义。
(2)state
state的类型指定了如果将storm的中间输出或者最终输出持久化到某个地方(如内存),当某个批次的数据重放时应该如果更新状态。state对于下游出现错误的情况尤其有用。
事务型状态:同一批次tuple提供的结果是相同的。
非事务型状态:没有回滚能力,更新操作是永久的。
透明事务型状态:更新操作基于先前的值,这样由于这批数据发生变化,对应的结果也会发生变化。透明事务型状态除了保存当前数据外,还要保存上一批数据,当数据重放时,可以基于上一批数据作更新。
(二)看官方提供的示例
package org.ljh.tridentdemo; import backtype.storm.Config; import backtype.storm.LocalCluster; import backtype.storm.LocalDRPC; import backtype.storm.StormSubmitter; import backtype.storm.generated.StormTopology; import backtype.storm.tuple.Fields; import backtype.storm.tuple.Values; import storm.trident.TridentState; import storm.trident.TridentTopology; import storm.trident.operation.BaseFunction; import storm.trident.operation.TridentCollector; import storm.trident.operation.builtin.Count; import storm.trident.operation.builtin.FilterNull; import storm.trident.operation.builtin.MapGet; import storm.trident.operation.builtin.Sum; import storm.trident.testing.FixedBatchSpout; import storm.trident.testing.MemoryMapState; import storm.trident.tuple.TridentTuple; public class TridentWordCount { public static class Split extends BaseFunction { @Override public void execute(TridentTuple tuple, TridentCollector collector) { String sentence = tuple.getString(0); for (String word : sentence.split(" ")) { collector.emit(new Values(word)); } } } public static StormTopology buildTopology(LocalDRPC drpc) { FixedBatchSpout spout = new FixedBatchSpout(new Fields("sentence"), 3, new Values( "the cow jumped over the moon"), new Values( "the man went to the store and bought some candy"), new Values( "four score and seven years ago"), new Values("how many apples can you eat"), new Values( "to be or not to be the person")); spout.setCycle(true); //创建拓扑对象 TridentTopology topology = new TridentTopology(); //这个流程用于统计单词数据,结果将被保存在wordCounts中 TridentState wordCounts = topology.newStream("spout1", spout) .parallelismHint(16) .each(new Fields("sentence"), new Split(), new Fields("word")) .groupBy(new Fields("word")) .persistentAggregate(new MemoryMapState.Factory(), new Count(), new Fields("count")).parallelismHint(16); //这个流程用于查询上面的统计结果 topology.newDRPCStream("words", drpc) .each(new Fields("args"), new Split(), new Fields("word")) .groupBy(new Fields("word")) .stateQuery(wordCounts, new Fields("word"), new MapGet(), new Fields("count")) .each(new Fields("count"), new FilterNull()) .aggregate(new Fields("count"), new Sum(), new Fields("sum")); return topology.build(); } public static void main(String[] args) throws Exception { Config conf = new Config(); conf.setMaxSpoutPending(20); if (args.length == 0) { LocalDRPC drpc = new LocalDRPC(); LocalCluster cluster = new LocalCluster(); cluster.submitTopology("wordCounter", conf, buildTopology(drpc)); for (int i = 0; i < 100; i++) { System.out.println("DRPC RESULT: " + drpc.execute("words", "cat the dog jumped")); Thread.sleep(1000); } } else { conf.setNumWorkers(3); StormSubmitter.submitTopologyWithProgressBar(args[0], conf, buildTopology(null)); } } }
实例实现了最基本的wordcount功能,然后将结果输出。关键步骤如下:
1、定义了输入流
FixedBatchSpout spout = new FixedBatchSpout(new Fields("sentence"), 3, new Values( "the cow jumped over the moon"), new Values( "the man went to the store and bought some candy"), new Values( "four score and seven years ago"), new Values("how many apples can you eat"), new Values( "to be or not to be the person")); spout.setCycle(true);
(1)使用FixedBatchSpout创建一个输入spout,spout的输出字段为sentence,每3个元组作为一个batch。
(2)数据不断的重复发送。
2、统计单词数量
TridentState wordCounts = topology.newStream("spout1", spout) .parallelismHint(16) .each(new Fields("sentence"), new Split(), new Fields("word")) .groupBy(new Fields("word")) .persistentAggregate(new MemoryMapState.Factory(), new Count(), new Fields("count")).parallelismHint(16);
这个流程用于统计单词数据,结果将被保存在wordCounts中。6行代码的含义分别为:
(1)首先从spout中读取消息,spout1定义了zookeeper中用于保存这个拓扑的节点名称。
(2)并行度设置为16,即16个线程同时从spout中读取消息。
(3)each中的三个参数分别为:输入字段名称,处理函数,输出字段名称。即从字段名称叫sentence的数据流中读取数据,然后经过new Split()处理后,以word作为字段名发送出去。其中new Split()后面介绍,它的功能就是将输入的内容以空格为界作了切分。
(4)将字段名称为word的数据流作分组,即相同值的放在一组。
(5)将已经分好组的数据作统计,结果放到MemoryMapState,然后以count作为字段名称将结果发送出去。这步骤会同时存储数据及状态,并将返回TridentState对象。
(6)并行度设置。
3、输出统计结果
topology.newDRPCStream("words", drpc) .each(new Fields("args"), new Split(), new Fields("word")) .groupBy(new Fields("word")) .stateQuery(wordCounts, new Fields("word"), new MapGet(), new Fields("count")) .each(new Fields("count"), new FilterNull()) .aggregate(new Fields("count"), new Sum(), new Fields("sum"));
这个流程从上述的wordCounts对象中读取结果,并返回。6行代码的含义分别为:
(1)等待一个drpc调用,从drpc服务器中接受words的调用来提供消息。调用代码如下:
drpc.execute("words", "cat the dog jumped")
(2)输入为上述调用中提供的参数,经过Split()后,以word作为字段名称发送出去。
(3)以word的值作分组。
(4)从wordCounts对象中查询结果。4个参数分别代表:数据来源,输入数据,内置方法(用于从map中根据key来查找value),输出名称。
(5)过滤掉空的查询结果,如本例中,cat和dog都没有结果。
(6)将结果作统计,并以sum作为字段名称发送出去,这也是DRPC调用所返回的结果。如果没有这一行,最后的输出结果
DRPC RESULT: [["cat the dog jumped","the",2310],["cat the dog jumped","jumped",462]]
加上这一行后,结果为:
DRPC RESULT: [[180]]
4、split的字义
public static class Split extends BaseFunction { @Override public void execute(TridentTuple tuple, TridentCollector collector) { String sentence = tuple.getString(0); for (String word : sentence.split(" ")) { collector.emit(new Values(word)); } } }
注意它最后会发送数据。
5、创建并启动拓扑
public static void main(String[] args) throws Exception { Config conf = new Config(); conf.setMaxSpoutPending(20); if (args.length == 0) { LocalDRPC drpc = new LocalDRPC(); LocalCluster cluster = new LocalCluster(); cluster.submitTopology("wordCounter", conf, buildTopology(drpc)); for (int i = 0; i < 100; i++) { System.out.println("DRPC RESULT: " + drpc.execute("words", "cat the dog jumped")); Thread.sleep(1000); } } else { conf.setNumWorkers(3); StormSubmitter.submitTopologyWithProgressBar(args[0], conf, buildTopology(null)); } }
(1)当无参数运行时,启动一个本地的集群,及自已创建一个drpc对象来输入。
(2)当有参数运行时,设置worker数量为3,然后提交拓扑到集群,并等待远程的drpc调用。
(三)使用kafka作为数据源的一个例子
package com.netease.sytopology; import java.io.File; import java.io.FileWriter; import java.io.IOException; import java.util.Arrays; import org.slf4j.Logger; import org.slf4j.LoggerFactory; import storm.kafka.BrokerHosts; import storm.kafka.StringScheme; import storm.kafka.ZkHosts; import storm.kafka.trident.OpaqueTridentKafkaSpout; import storm.kafka.trident.TridentKafkaConfig; import storm.trident.TridentTopology; import storm.trident.operation.BaseFunction; import storm.trident.operation.TridentCollector; import storm.trident.operation.builtin.Count; import storm.trident.testing.MemoryMapState; import storm.trident.tuple.TridentTuple; import backtype.storm.Config; import backtype.storm.StormSubmitter; import backtype.storm.generated.AlreadyAliveException; import backtype.storm.generated.InvalidTopologyException; import backtype.storm.generated.StormTopology; import backtype.storm.spout.SchemeAsMultiScheme; import backtype.storm.tuple.Fields; import backtype.storm.tuple.Values; /* * 本类完成以下内容 */ public class SyTopology { public static final Logger LOG = LoggerFactory.getLogger(SyTopology.class); private final BrokerHosts brokerHosts; public SyTopology(String kafkaZookeeper) { brokerHosts = new ZkHosts(kafkaZookeeper); } public StormTopology buildTopology() { TridentKafkaConfig kafkaConfig = new TridentKafkaConfig(brokerHosts, "ma30", "storm"); kafkaConfig.scheme = new SchemeAsMultiScheme(new StringScheme()); // TransactionalTridentKafkaSpout kafkaSpout = new // TransactionalTridentKafkaSpout(kafkaConfig); OpaqueTridentKafkaSpout kafkaSpout = new OpaqueTridentKafkaSpout(kafkaConfig); TridentTopology topology = new TridentTopology(); // TridentState wordCounts = topology.newStream("kafka4", kafkaSpout). each(new Fields("str"), new Split(), new Fields("word")).groupBy(new Fields("word")) .persistentAggregate(new MemoryMapState.Factory(), new Count(), new Fields("count")).parallelismHint(16); // .persistentAggregate(new HazelCastStateFactory(), new Count(), // new Fields("aggregates_words")).parallelismHint(2); return topology.build(); } public static void main(String[] args) throws AlreadyAliveException, InvalidTopologyException { String kafkaZk = args[0]; SyTopology topology = new SyTopology(kafkaZk); Config config = new Config(); config.put(Config.TOPOLOGY_TRIDENT_BATCH_EMIT_INTERVAL_MILLIS, 2000); String name = args[1]; String dockerIp = args[2]; config.setNumWorkers(9); config.setMaxTaskParallelism(5); config.put(Config.NIMBUS_HOST, dockerIp); config.put(Config.NIMBUS_THRIFT_PORT, 6627); config.put(Config.STORM_ZOOKEEPER_PORT, 2181); config.put(Config.STORM_ZOOKEEPER_SERVERS, Arrays.asList(dockerIp)); StormSubmitter.submitTopology(name, config, topology.buildTopology()); } static class Split extends BaseFunction { public void execute(TridentTuple tuple, TridentCollector collector) { String sentence = tuple.getString(0); for (String word : sentence.split(",")) { try { FileWriter fw = new FileWriter(new File("/home/data/test/ma30/ma30.txt"),true); fw.write(word); fw.flush(); fw.close(); } catch (IOException e) { // TODO Auto-generated catch block e.printStackTrace(); } collector.emit(new Values(word)); } } } }
本例将从kafka中读取消息,然后对消息根据“,”作拆分,并写入一个本地文件。
1、定义kafka想着配置
TridentKafkaConfig kafkaConfig = new TridentKafkaConfig(brokerHosts, "ma30", "storm");
kafkaConfig.scheme = new SchemeAsMultiScheme(new StringScheme());
OpaqueTridentKafkaSpout kafkaSpout = new OpaqueTridentKafkaSpout(kafkaConfig);
其中ma30是订阅的topic名称。
2、从kafka中读取消息并处理
topology.newStream("kafka4", kafkaSpout).
each(new Fields("str"), new Split(),new Fields("word")).
groupBy(new Fields("word"))
.persistentAggregate(new MemoryMapState.Factory(), new Count(),
new Fields("count")).parallelismHint(16);
(1)指定了数据来源,并指定zookeeper中用于保存数据的位置,即保存在/transactional/kafka4。
(2)指定处理方法及发射的字段。
(3)根据word作分组。
(4)计数后将状态写入MemoryMapState
提交拓扑:
storm jar target/sytopology2-0.0.1-SNAPSHOT.jar com.netease.sytopology.SyTopology 192.168.172.98:2181/kafka test3 192.168.172.98
此时可以在/home/data/test/ma30/ma30.txt看到split的结果