数据流分组定义了一个数据流中的tuple如何分发给topology中不同bolt的task。
Shuffle grouping(随机分组):这种方式会随机分发 tuple 给 bolt 的各个 task,每个bolt 实例接收到的相同数量的 tuple。
Fields grouping(按字段分组):根据指定字段的值进行分组。比如说,一个数据流根据“word”字段进行分组,所有具有相同“word”字段值的 tuple 会路由到同一个 bolt 的 task 中。
All grouping(全复制分组):将所有的 tuple 复制后分发给所有 bolt task。每个订阅数据流的 task 都会接收到 tuple 的拷贝。
Globle grouping(全局分组):这种分组方式将所有的 tuples 路由到唯一一个 task 上。Storm 按照最小的 task ID 来选取接收数据的 task。注意,当使用全局分组方式时,设置 bolt 的 task 并发度是没有意义的,因为所有 tuple 都转发到同一个 task 上了。使用全局分组的时候需要注意,因为所有的 tuple 都转发到一个 JVM 实例上,可能会引起 Storm 集群中某个 JVM 或者服务器出现性能瓶颈或崩溃。
None grouping(不分组):在功能上和随机分组相同,是为将来预留的。
Direct grouping(指向型分组):数据源会调用 emitDirect() 方法来判断一个 tuple 应该由哪个 Storm 组件来接收。只能在声明了是指向型的数据流上使用。
Local or shuffle grouping(本地或随机分组):和随机分组类似,但是,会将 tuple 分发给同一个 worker 内的 bolt task(如果 worker 内有接收数据的 bolt task)。其他情况下,采用随机分组的方式。取决于 topology 的并发度,本地或随机分组可以减少网络传输,从而提高 topology 性能。
随机分组
最经常用的也就是Shuffle grouping,Fields grouping,Direct grouping等等
现在我们看一个例子:
就是最经常见的数单词的例子
- public class WordCountBolt extends BaseRichBolt{
- private OutputCollector collector;
- private HashMap<String, Long> counts = null;
-
-
- public void prepare(Map config, TopologyContext context,
- OutputCollector collector) {
- this.collector = collector;
- this.counts = new HashMap<String, Long>();
- }
-
-
- public void execute(Tuple tuple) {
- String word = tuple.getStringByField("word");
- Long count = this.counts.get(word);
- if(count == null){
- count = 0L;
- }
- count++;
- this.counts.put(word, count);
- this.collector.emit(new Values(word, count));
- }
-
-
- public void declareOutputFields(OutputFieldsDeclarer declarer) {
- declarer.declare(new Fields("word", "count"));
- }
- }
在添加这个bolt的时候,使用的是按字段分组,如下
- builder.setBolt(COUNT_BOLT_ID, countBolt,4)
- .fieldsGrouping(SPLIT_BOLT_ID, new Fields("word"));
如果我们分组模式改成
- builder.setBolt(COUNT_BOLT_ID, countBolt,4)
- .shuffleGrouping(SPLIT_BOLT_ID);
那么对单词的统计就会偏少。
为什么?
大家想想恩,有4个countbolt实例(咱们暂时称之为countbolta,b,c,d),如果我是随机分组,the这个单词出现了3回,前两回被分配到了countbolta,第三回被分配到了countboltb,那么后面的reportbolt先收到了<the,2>这个tuple(来自countbolta),然后又收到了<the,1>这个tuple(来自countboltb),最后的输出肯定是the:1喽
那么如果使用
- builder.setBolt(COUNT_BOLT_ID, countBolt,4)
- .fieldsGrouping(SPLIT_BOLT_ID, new Fields("word"));
自然就不会出现刚才的问题了,为什么,大家自己想。
直接分组
这里我引用一个用storm给句子加感叹号的例子,代码在最后
////////////////////////////////////////////////////////////////////////////////////////////////////////////////
以下为16-7-4修改
其实我下面这例子不好
直接分组,主要是保证把消息给bolt中某一个特定的task
而下面的例子的实际效果是想吧 messagea给bolta,messageb给boltb
那么其实还有更方便的做法,就是
在发送是:
public void execute(Tuple tuple, BasicOutputCollector collector) {
tpsCounter.count();
Long tupleId = tuple.getLong(0);
Object obj = tuple.getValue(1);
if (obj instanceof TradeCustomer) {
TradeCustomer tradeCustomer = (TradeCustomer)obj;
Pair trade = tradeCustomer.getTrade();
Pair customer = tradeCustomer.getCustomer();
collector.emit(SequenceTopologyDef.TRADE_STREAM_ID,
new Values(tupleId, trade));
collector.emit(SequenceTopologyDef.CUSTOMER_STREAM_ID,
new Values(tupleId, customer));
}else if (obj != null){
LOG.info("Unknow type " + obj.getClass().getName());
}else {
LOG.info("Nullpointer " );
}
}
在提交时:
builder.setBolt(SequenceTopologyDef.SPLIT_BOLT_NAME, new SplitRecord(), 2).shuffleGrouping(
SequenceTopologyDef.SEQUENCE_SPOUT_NAME);
builder.setBolt(SequenceTopologyDef.TRADE_BOLT_NAME, new PairCount(), 1).shuffleGrouping(
SequenceTopologyDef.SPLIT_BOLT_NAME, // --- 发送方名字
SequenceTopologyDef.TRADE_STREAM_ID); // --- 接收发送方该stream 的tuple
builder.setBolt(SequenceTopologyDef.CUSTOMER_BOLT_NAME, new PairCount(), 1)
.shuffleGrouping(SequenceTopologyDef.SPLIT_BOLT_NAME, // --- 发送方名字
SequenceTopologyDef.CUSTOMER_STREAM_ID); // --- 接收发送方该stream 的tuple
定义输出格式
public void declareOutputFields(OutputFieldsDeclarer declarer) {
declarer.declareStream(SequenceTopologyDef.TRADE_STREAM_ID, new Fields("ID", "TRADE"));
declarer.declareStream(SequenceTopologyDef.CUSTOMER_STREAM_ID, new Fields("ID", "CUSTOMER"));
}
最后接收的时候还得判断一下
if (input.getSourceStreamId().equals(SequenceTopologyDef.TRADE_STREAM_ID) ) {
customer = pair;
customerTuple = input;
tradeTuple = tradeMap.get(tupleId);
if (tradeTuple == null) {
customerMap.put(tupleId, input);
return;
}
trade = (Pair) tradeTuple.getValue(1);
}
参考资料
数据的分流
以上为16-7-4修改
////////////////////////////////////////////////////////////////////////////////////////////////////////////////
最开始的时候
运行的结果如下:
- mystorm.PrintBolt@67178f5d String recieved: edi:I'm happy!
- mystorm.PrintBolt@67178f5d String recieved: marry:I'm angry!
- mystorm.PrintBolt@393ddf54 String recieved: ted:I'm excited!
- mystorm.PrintBolt@393ddf54 String recieved: john:I'm sad!
- mystorm.PrintBolt@5f97cfcb String recieved: marry:I'm angry!
不同的task都平均收到了tuple
然后我想让指定某些句子只让某个task接受,怎么办?
首先看ExclaimBasicBolt
- public class ExclaimBasicBolt extends BaseBasicBolt {
-
-
-
-
-
- private static final long serialVersionUID = -6239845315934660303L;
- private List<Integer> list;
- private List<Integer> list2;
-
-
- @Override
- public void execute(Tuple tuple, BasicOutputCollector collector) {
-
- String sentence = (String) tuple.getValue(0);
- String out = sentence + "!";
- if (out.startsWith("e")) {
- collector.emitDirect(list.get(0),new Values(out));
- }else {
- collector.emitDirect(list2.get(0),new Values(out));
- }
-
- }
-
-
- @Override
- public void declareOutputFields(OutputFieldsDeclarer declarer) {
- declarer.declare(true,new Fields("excl_sentence"));
- }
-
-
- @Override
- public void prepare(Map stormConf, TopologyContext context) {
- list =context.getComponentTasks("print");
- list2=context.getComponentTasks("print2");
- }
- }
在构建topology的时候
使用directGrouping
- builder.setSpout("spout", new RandomSpout());
- builder.setBolt("exclaim", new ExclaimBasicBolt(),3).shuffleGrouping("spout");
- builder.setBolt("print", new PrintBolt(),3).directGrouping("exclaim");
- builder.setBolt("print2", new PrintBolt2(),3).directGrouping("exclaim");
PrintBolt2与PrintBolt类似
只是打印的时候打印出 System.err.println(this+" i am two String recieved: " + rec);
OK这下运行的时候我们就能看到
- mystorm.PrintBolt2@238ac8bf String recieved: ted:I'm excited!
- mystorm.PrintBolt2@238ac8bf String recieved: john:I'm sad!
- mystorm.PrintBolt2@238ac8bf String recieved: marry:I'm angry!
- mystorm.PrintBolt2@238ac8bf String recieved: ted:I'm excited!
- mystorm.PrintBolt@611b7a20 i am two String recieved: edi:I'm happy!
- mystorm.PrintBolt@611b7a20 i am two String recieved: edi:I'm happy!
- mystorm.PrintBolt@611b7a20 i am two String recieved: edi:I'm happy!
- mystorm.PrintBolt@611b7a20 i am two String recieved: edi:I'm happy!
- mystorm.PrintBolt@611b7a20 i am two String recieved: edi:I'm happy!
- mystorm.PrintBolt@611b7a20 i am two String recieved: edi:I'm happy!
- mystorm.PrintBolt@611b7a20 i am two String recieved: edi:I'm happy!
- mystorm.PrintBolt@611b7a20 i am two String recieved: edi:I'm happy!
- mystorm.PrintBolt@611b7a20 i am two String recieved: edi:I'm happy!
- mystorm.PrintBolt2@238ac8bf String recieved: marry:I'm angry!
- mystorm.PrintBolt2@238ac8bf String recieved: ted:I'm excited!
- mystorm.PrintBolt2@238ac8bf String recieved: marry:I'm angry!
所有e开头的句子 都跑到Print2这个Bolt的某个task里面了。
本节的整体代码见
- package mystorm;
-
- public class ExclaimBasicTopo {
- public static void main(String[] args) throws Exception {
- TopologyBuilder builder = new TopologyBuilder();
-
- builder.setSpout("spout", new RandomSpout());
- builder.setBolt("exclaim", new ExclaimBasicBolt(),3).shuffleGrouping("spout");
- builder.setBolt("print", new PrintBolt(),3).shuffleGrouping("exclaim");
-
- Config conf = new Config();
- conf.setDebug(false);
-
- if (args != null && args.length > 0) {
- conf.setNumWorkers(3);
- StormSubmitter.submitTopology(args[0], conf, builder.createTopology());
- } else {
- LocalCluster cluster = new LocalCluster();
- cluster.submitTopology("test", conf, builder.createTopology());
- }
- }
- }
-
-
- package mystorm;
-
- public class RandomSpout extends BaseRichSpout {
-
- private SpoutOutputCollector collector;
- private Random rand;
- private int index;
- private static String[] sentences = new String[] {
- "edi:I'm happy", "marry:I'm angry", "john:I'm sad", "ted:I'm excited", "laden:I'm dangerous"};
-
- @Override
- public void open(Map conf, TopologyContext context,SpoutOutputCollector collector) {
- this.collector = collector;
- this.rand = new Random();
- }
-
- @Override
- public void nextTuple() {
- if (index<10*sentences.length) {
- String toSay = sentences[rand.nextInt(sentences.length)];
- this.collector.emit(new Values(toSay));
- index++;
- }else {
- try {
- Thread.sleep(1000);
- System.out.println("我停了一秒");
- } catch (InterruptedException e) {
-
- e.printStackTrace();
- }
- }
-
- }
-
- @Override
- public void declareOutputFields(OutputFieldsDeclarer declarer) {
- declarer.declare(new Fields("sentence"));
- }
-
- }
-
- package mystorm;
-
- public class ExclaimBasicBolt extends BaseBasicBolt {
-
- @Override
- public void execute(Tuple tuple, BasicOutputCollector collector) {
-
- String sentence = (String) tuple.getValue(0);
- String out = sentence + "!";
- collector.emit(new Values(out));
- }
-
- @Override
- public void declareOutputFields(OutputFieldsDeclarer declarer) {
- declarer.declare(new Fields("excl_sentence"));
- }
-
- }
-
-
- package mystorm;
-
- public class PrintBolt extends BaseBasicBolt {
-
- @Override
- public void execute(Tuple tuple, BasicOutputCollector collector) {
- String rec = tuple.getString(0);
- System.err.println(this+" String recieved: " + rec);
- }
- @Override
- public void declareOutputFields(OutputFieldsDeclarer declarer) {
-
- }
-
- }