Create real-time stream processing applications with Apache Storm

Learning Storm

Create real-time stream processing applications with Apache Storm

Ankit Jain Anand Nalya

BIRMINGHAM - MUMBAI

Copyright © 2014 Packt Publishing

All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, without the prior written permission of the publisher, except in the case of brief quotations embedded in critical articles or reviews.

Every effort has been made in the preparation of this book to ensure the accuracy of the information presented. However, the information contained in this book is sold without warranty, either express or implied. Neither the authors, nor Packt Publishing, and its dealers and distributors will be held liable for any damages caused or alleged to be caused directly or indirectly by this book.

Packt Publishing has endeavored to provide trademark information about all of the companies and products mentioned in this book by the appropriate use of capitals.

However, Packt Publishing cannot guarantee the accuracy of this information.

First published: August 2014

Production reference: 1200814

Published by Packt Publishing Ltd.

Livery Place 35 Livery Street

Birmingham B3 2PB, UK.

ISBN 978-1-78398-132-8 www.packtpub.com

Cover image by Pratyush Mohanta ( [email protected] )

Credits

Authors Ankit Jain Anand Nalya

Reviewers Vinoth Kannan Sonal Raj

Danijel Schiavuzzi

Commissioning Editor Usha Iyer

Acquisition Editor Llewellyn Rozario

Content Development Editor Sankalp Pawar

Technical Editors Menza Mathew Siddhi Rane

Copy Editors Sarang Chari Mradula Hegde

Project Coordinator Harshal Ved

Proofreaders Simran Bhogal Ameesha Green Paul Hindle

Indexers Hemangini Bari Tejal Soni Priya Subramani

Graphics Abhinash Sahu

Production Coordinator Saiprasad Kadam

Cover Work

Saiprasad Kadam

About the Authors

Ankit Jain holds a Bachelor's degree in Computer Science Engineering. He has 4 years of experience in designing and architecting solutions for the Big Data domain and has been involved with several complex engagements. His technical strengths include Hadoop, Storm, S4, HBase, Hive, Sqoop, Flume, ElasticSearch, Machine Learning, Kafka, Spring, Java, and J2EE. He is currently employed with Impetus Infotech Pvt. Ltd.

He also shares his thoughts on his personal blog at http://ankitasblogger.

blogspot.in/ . You can follow him on Twitter at @mynameisanky . He spends most of his time reading books and playing with different technologies. When not at work, he spends time with his family and friends watching movies and playing games.

I would like to thank my family and colleagues for always being

there for me. Special thanks to the Packt Publishing team; without

you guys, this work would not have been possible.

in designing, developing, deploying, and benchmarking Big Data and web-scale applications for both start-ups and enterprises. He focuses on reducing the complexity in getting things done with brevity in code.

He blogs about Big Data, web applications, and technology in general at

http://anandnalya.com/ . You can also follow him on Twitter at @anandnalya . When not working on projects, he can be found stargazing or reading.

I would like to thank my wife, Nidhi, for putting up with so many

of my side projects and my family members who are always there

for me. Special thanks to my colleagues who helped me validate the

writing, and finally, the reviewers and editors at Packt Publishing,

without whom this work would not have been possible.

About the Reviewers

Vinoth Kannan is a solution architect at WidasConcepts, Germany, that focuses on creating robust, highly scalable, real-time systems for storage, search, and analytics. He now works in Germany after his professional stints in France, Italy, and India.

Currently, he works extensively with open source frameworks based on Storm, Hadoop, and NoSQL databases. He has helped design and develop complex, real-time Big Data systems for some of the largest financial institutions and e-commerce companies.

He also co-organizes the Big Data User group in Karlsruhe and Stuttgart in Germany, and is a regular speaker at user group meets and international conferences on Big Data. He holds a double Master's degree in Communication Systems Engineering from Politecnico di Torino, Italy, and Grenoble Institute of Technology, France.

This is for my wonderful parents and my beloved wife, Sudha.

engineer with dreams. He has been a research fellow at SERC, IISc, Bangalore, and he has pursued projects on distributed computing and real-time operations. He has spoken at PyCon India on Storm and Neo4J and has published articles and research papers in leading magazines and international journals. Presently, he works at Sigmoid Analytics, where he is actively involved in the development of machine-learning frameworks and Big Data solutions.

I am grateful to Ankit and Anand for patiently listening to my critiques, and I'd like to thank the open source community for keeping their passion alive and contributing to remarkable projects such as Storm. A special thank you to my parents, without whom I never would have grown to love learning as much as I do.

Danijel Schiavuzzi is a software engineer and technology enthusiast with a passionate interest in systems programming and distributed systems.

Currently, he works at Infobip, where he finds new usages for Storm and other Big Data technologies in the telecom domain on a daily basis. He has a strong focus on real-time data analytics, log processing, and external systems monitoring and alerting. He is passionate about open source, having contributed a few minor patches to Storm itself.

In his spare time, he enjoys reading a book, following space exploration and scientific and technological news, tinkering with various gadgets, listening and occasionally playing music, discovering old art movie masterpieces, and enjoying cycling around beautiful natural sceneries.

I would like to thank the Apache Storm community for developing

such a great technology and making distributed computing more fun.

www.PacktPub.com

Support files, eBooks, discount offers, and more

You might want to visit www.PacktPub.com for support files and downloads related to your book.

Did you know that Packt offers eBook versions of every book published, with PDF and ePub files available? You can upgrade to the eBook version at www.PacktPub.com and as a print book customer, you are entitled to a discount on the eBook copy. Get in touch with us at [email protected] for more details.

At www.PacktPub.com, you can also read a collection of free technical articles, sign up for a range of free newsletters, and receive exclusive discounts and offers on Packt books and eBooks.

TM

http://PacktLib.PacktPub.com

Do you need instant solutions to your IT questions? PacktLib is Packt's online digital book library. Here, you can access, read, and search across Packt's entire library of books.

Why subscribe?

• Fully searchable across every book published by Packt

• Copy and paste, print, and bookmark content

• On demand and accessible via web browser

Free access for Packt account holders

If you have an account with Packt at www.PacktPub.com, you can use this to access PacktLib

today and view nine entirely free books. Simply use your login credentials for immediate access.

Table of Contents

Preface 1 Chapter 1: Setting Up Storm on a Single Machine 7

Features of Storm 8

Storm components 9

Nimbus 9

Supervisor nodes 9

The ZooKeeper cluster 10

The Storm data model 10

Definition of a Storm topology 11

Operation modes 14

Setting up your development environment 15

Installing Java SDK 6 15

Installing Maven 16

Installing Git – distributed version control 17

Installing the STS IDE 17

Developing a sample topology 19

Setting up ZooKeeper 25

Setting up Storm on a single development machine 26 Deploying the sample topology on a single-node cluster 28 Summary 31 Chapter 2: Setting Up a Storm Cluster 33

Setting up a ZooKeeper cluster 33

Setting up a distributed Storm cluster 37 Deploying a topology on a remote Storm cluster 39 Deploying the sample topology on the remote cluster 40 Configuring the parallelism of a topology 42

The worker process 42

Tasks 42 Configuring parallelism at the code level 43 Distributing worker processes, executors, and tasks in the

sample topology 44

Rebalancing the parallelism of a topology 45 Rebalancing the parallelism of the sample topology 46

Stream grouping 48

Shuffle grouping 48

Fields grouping 48

All grouping 49

Global grouping 50

Direct grouping 50

Local or shuffle grouping 51

Custom grouping 52

Guaranteed message processing 53

Summary 55

Chapter 3: Monitoring the Storm Cluster 57

Starting to use the Storm UI 57

Monitoring a topology using the Storm UI 58 Cluster statistics using the Nimbus thrift client 65 Fetching information with the Nimbus thrift client 65 Summary 78 Chapter 4: Storm and Kafka Integration 79

The Kafka architecture 80

The producer 80

Replication 81 Consumers 81 Brokers 82

Data retention 83

Setting up Kafka 83

Setting up a single-node Kafka cluster 83 Setting up a three-node Kafka cluster 86

Running multiple Kafka brokers on a single node 88

A sample Kafka producer 89

Integrating Kafka with Storm 92

Summary 98 Chapter 5: Exploring High-level Abstraction in Storm with Trident 99

Introducing Trident 100

Understanding Trident's data model 100

Writing Trident functions, filters, and projections 100

Trident functions 101

Trident filters 102

Trident projections 103

Trident repartitioning operations 104

The shuffle operation 104

The partitionBy operation 105

The global operation 106

The broadcast operation 107

The batchGlobal operation 108

The partition operation 108

Trident aggregators 109

The partition aggregate 110

The aggregate 110

The ReducerAggregator interface 111

The Aggregator interface 112

The CombinerAggregator interface 113

The persistent aggregate 114

Aggregator chaining 114

Utilizing the groupBy operation 115

A non-transactional topology 116

A sample Trident topology 118

Maintaining the topology state with Trident 123

A transactional topology 124

The opaque transactional topology 125

Distributed RPC 126

When to use Trident 130

Summary 130 Chapter 6: Integration of Storm with Batch Processing Tools 131

Exploring Apache Hadoop 131

Understanding HDFS 132

Understanding YARN 134

Installing Apache Hadoop 135

Setting up password-less SSH 136

Getting the Hadoop bundle and setting up environment variables 137

Setting up HDFS 138

Setting up YARN 141

Integration of Storm with Hadoop 144

Setting up Storm-YARN 145

Deploying Storm-Starter topologies on Storm-YARN 149

Chapter 7: Integrating Storm with JMX, Ganglia, HBase,

and Redis 153

Monitoring the Storm cluster using JMX 154 Monitoring the Storm cluster using Ganglia 156

Integrating Storm with HBase 166

Integrating Storm with Redis 177

Summary 182 Chapter 8: Log Processing with Storm 183

Server log-processing elements 183

Producing the Apache log in Kafka 184

Splitting the server log line 188

Identifying the country, the operating system type,

and the browser type from the logfile 192

Extracting the searched keyword 196

Persisting the process data 198

Defining a topology and the Kafka spout 204

Deploying a topology 208

MySQL queries 209

Calculating the page hits from each country 209 Calculating the count for each browser 211 Calculating the count for each operating system 211 Summary 211

Chapter 9: Machine Learning 213

Exploring machine learning 213

Using Trident-ML 214

The use case – clustering synthetic control data 216

Producing a training dataset into Kafka 216

Building a Trident topology to build the clustering model 220

Summary 227

Index 229

Preface

Real-time data processing is no longer a luxury exercised by a few big companies but has become a necessity for businesses that want to compete, and Apache Storm is becoming the de facto standard to develop real-time processing pipelines. The key features of Storm are that it is horizontally scalable, fault-tolerant, and provides guaranteed message processing. Storm can solve various types of analytical problems, such as machine learning, log processing, and graph analysis.

Learning Storm will serve both as a getting-started guide for inexperienced developers and as a reference to implement advanced use cases with Storm for experienced developers. In the first two chapters, you will learn the basics of a Storm topology and various components of a Storm cluster. In the later chapters, you will learn how to build a Storm application that can interact with various other Big Data technologies and how to create transactional topologies. Finally, the last two chapters cover case studies for log processing and machine learning.

What this book covers

Chapter 1, Setting Up Storm on a Single Machine, gives an introduction to Storm and its components, followed by setting up a single-node Storm cluster, developing a sample Storm topology, and deploying it on a single-node cluster.

Chapter 2, Setting Up a Storm Cluster, covers the deployment of Storm in the cluster, deploys sample topology on a Storm cluster, discusses how we can achieve parallelism in Storm and how we can change the parallelism of the Storm topology in runtime, and even covers the basic Storm commands.

Chapter 3, Monitoring the Storm Cluster, introduces you to various ways of monitoring

a Storm cluster, including the Storm UI and the Nimbus thrift client.

Chapter 4, Storm and Kafka Integration, introduces Apache Kafka, a message-queuing system, and shows how to integrate it with Storm to interact with data coming from external systems.

Chapter 5, Exploring High-level Abstraction in Storm with Trident, gives an introduction to Trident's function, filter, projection, aggregator, and repartitioning operations.

It also covers a description of the transactional, non-transactional, and opaque transactional topologies. At the end, we cover how we can develop the sample Trident topology and how we can use the distributed RPC feature.

Chapter 6, Integration of Storm with Batch Processing Tools, shows you how to integrate Storm with Hadoop using the Storm-YARN framework.

Chapter 7, Integrating Storm with JMX, Ganglia, HBase, and Redis, shows you how to integrate Storm with various other Big Data technologies. It also focuses on how we can publish Storm's JVM metrics on Ganglia.

Chapter 8, Log Processing with Storm, covers a sample log processing application in which, we parse Apache web server logs and generate some business information from logfiles.

Chapter 9, Machine Learning, walks you through a case study of implementing a machine learning topology in Storm.

What you need for this book

All of the code in this book has been tested on CentOS 6.4. It will run on other variants of Linux and Windows as well with respective changes in commands.

We have tried to keep the chapters self-contained, and the setup and installation of all the software used in each chapter is included in the chapter itself. The following software packages are used throughout the book:

• CentOS 6.4

• Oracle JDK 6/7

• Apache ZooKeeper 3.4.5

• Apache Storm 0.9.0.1

• Eclipse or Spring Tool Suite

Who this book is for

If you are a Java developer who wants to enter the world of real-time stream processing applications using Apache Storm, then this book is for you. No previous experience in Storm is required as this book starts from the basics. After finishing this book, you will be able to develop simple Storm applications.

Conventions

In this book, you will find a number of styles of text that distinguish between different kinds of information. Here are some examples of these styles and an explanation of their meaning.

Code words in text, database table names, folder names, filenames, file extensions, pathnames, dummy URLs, user input, and Twitter handles are shown as follows:

"The LearningStormBolt class extends the serialized BaseRichBolt class."

A block of code is set as follows:

public void open(Map conf, TopologyContext context, SpoutOutputCollector spoutOutputCollector) {

this.spoutOutputCollector = spoutOutputCollector;

}

Any command-line input or output is written as follows:

# bin/storm nimbus

New terms and important words are shown in bold. Words that you see on the screen, in menus or dialog boxes for example, appear in the text like this:

"Specify com.learningstorm as Group Id and storm-example as Artifact Id."

Warnings or important notes appear in a box like this.

Tips and tricks appear like this.

Reader feedback

Feedback from our readers is always welcome. Let us know what you think about this book—what you liked or may have disliked. Reader feedback is important for us to develop titles that you really get the most out of.

To send us general feedback, simply send an e-mail to [email protected] , and mention the book title via the subject of your message.

If there is a topic that you have expertise in and you are interested in either writing or contributing to a book, see our author guide on www.packtpub.com/authors .

Customer support

Now that you are the proud owner of a Packt book, we have a number of things to help you to get the most from your purchase.

Downloading the example code

You can download the example code files for all Packt books you have purchased from your account at http://www.packtpub.com . If you purchased this book elsewhere, you can visit http://www.packtpub.com/support and register to have the files e-mailed directly to you.

Errata

Although we have taken every care to ensure the accuracy of our content, mistakes do happen. If you find a mistake in one of our books—maybe a mistake in the text or the code—we would be grateful if you would report this to us. By doing so, you can save other readers from frustration and help us improve subsequent versions of this book. If you find any errata, please report them by visiting http://www.packtpub.

com/submit-errata , selecting your book, clicking on the errata submission form link,

and entering the details of your errata. Once your errata are verified, your submission

will be accepted and the errata will be uploaded on our website, or added to any list of

existing errata, under the Errata section of that title. Any existing errata can be viewed

by selecting your title from http://www.packtpub.com/support .

Piracy

Piracy of copyright material on the Internet is an ongoing problem across all media.

At Packt, we take the protection of our copyright and licenses very seriously. If you come across any illegal copies of our works, in any form, on the Internet, please provide us with the location address or website name immediately so that we can pursue a remedy.

Please contact us at [email protected] with a link to the suspected pirated material.

We appreciate your help in protecting our authors, and our ability to bring you valuable content.

Questions

You can contact us at [email protected] if you are having a problem with

any aspect of the book, and we will do our best to address it.

Setting Up Storm on a Single Machine

With the exponential growth in the amount of data being generated and advanced data-capturing capabilities, enterprises are facing the challenge of making sense out of this mountain of raw data. On the batch processing front, Hadoop has emerged as the go-to framework to deal with Big Data. Until recently, there has been a void when one looks for frameworks to build real-time stream processing applications.

Such applications have become an integral part of a lot of businesses as they enable them to respond swiftly to events and adapt to changing situations. Examples of this are monitoring social media to analyze public response to any new product that you launch and predicting the outcome of an election based on the sentiments of the election-related posts.

Apache Storm has emerged as the platform of choice for the industry leaders to develop such distributed, real-time, data processing platforms. It provides a set of primitives that can be used to develop applications that can process a very large amount of data in real time in a highly scalable manner.

Storm is to real-time processing what Hadoop is to batch processing. It is an open source software, currently being incubated at the Apache Software Foundation.

Being in incubation does not mean that it is not yet ready for actual production.

Indeed, it has been deployed to meet real-time processing needs by companies such as Twitter, Yahoo!, and Flipboard. Storm was first developed by Nathan Marz at BackType, a company that provided social search applications. Later, BackType was acquired by Twitter, and now it is a critical part of their infrastructure. Storm can be used for the following use cases:

• Stream processing: Storm is used to process a stream of data and update

a variety of databases in real time. This processing occurs in real time and

• Continuous computation: Storm can do continuous computation on data streams and stream the results into clients in real time. This might require processing each message as it comes or creating small batches over a little time. An example of continuous computation is streaming trending topics on Twitter into browsers.

• Distributed RPC: Storm can parallelize an intense query so that you can compute it in real time.

• Real-time analytics: Storm can analyze and respond to data that comes from different data sources as they happen in real time.

In this chapter, we will cover the following topics:

• Features of Storm

• Various components of a Storm cluster

• What is a Storm topology

• Local and remote operational modes to execute Storm topologies

• Setting up a development environment to develop a Storm topology

• Developing a sample topology

• Setting up a single-node Storm cluster and its prerequisites

• Deploying the sample topology

Features of Storm

The following are some of the features of Storm that make it a perfect solution to process streams of data in real time:

• Fast: Storm has been reported to process up to 1 million tuples per second per node.

• Horizontally scalable: Being fast is a necessary feature to build a high volume/velocity data processing platform, but a single-node will have an upper limit on the number of events that it can process per second. A node represents a single machine in your setup that execute Storm applications.

Storm, being a distributed platform, allows you to add more nodes to your

Storm cluster and increase the processing capacity of your application. Also,

it is linearly scalable, which means that you can double the processing capacity

by doubling the nodes.

• Fault tolerant: Units of work are executed by worker processes in a Storm cluster. When a worker dies, Storm will restart that worker, and if the node on which the worker is running dies, Storm will restart that worker on some other node in the cluster. The descriptions of the worker process is mentioned in the Configuring the parallelism of a topology section of Chapter 2, Setting Up a Storm Cluster.

• Guaranteed data processing: Storm provides strong guarantees that each message passed on to it to process will be processed at least once. In the event of failures, Storm will replay the lost tuples. Also, it can be configured so that each message will be processed only once.

• Easy to operate: Storm is simple to deploy and manage. Once the cluster is deployed, it requires little maintenance.

• Programming language agnostic: Even though the Storm platform runs on Java Virtual Machine, the applications that run over it can be written in any programming language that can read and write to standard input and output streams.

Storm components

A Storm cluster follows a master-slave model where the master and slave processes are coordinated through ZooKeeper. The following are the components of a Storm cluster.

Nimbus

The Nimbus node is the master in a Storm cluster. It is responsible for distributing the application code across various worker nodes, assigning tasks to different machines, monitoring tasks for any failures, and restarting them as and when required.

Nimbus is stateless and stores all of its data in ZooKeeper. There is a single Nimbus node in a Storm cluster. It is designed to be fail-fast, so when Nimbus dies, it can be restarted without having any effects on the already running tasks on the worker nodes. This is unlike Hadoop, where if the JobTracker dies, all the running jobs are left in an inconsistent state and need to be executed again.

Supervisor nodes

Supervisor nodes are the worker nodes in a Storm cluster. Each supervisor node runs a supervisor daemon that is responsible for creating, starting, and stopping worker processes to execute the tasks assigned to that node. Like Nimbus, a supervisor

daemon is also fail-fast and stores all of its state in ZooKeeper so that it can be restarted

The ZooKeeper cluster

In any distributed application, various processes need to coordinate with each other and share some configuration information. ZooKeeper is an application that provides all these services in a reliable manner. Being a distributed application, Storm also uses a ZooKeeper cluster to coordinate various processes. All of the states associated with the cluster and the various tasks submitted to the Storm are stored in ZooKeeper.

Nimbus and supervisor nodes do not communicate directly with each other but through ZooKeeper. As all data is stored in ZooKeeper, both Nimbus and the supervisor daemons can be killed abruptly without adversely affecting the cluster.

The following is an architecture diagram of a Storm cluster:

A Storm Cluster A Storm cluster's architecture

The Storm data model

The basic unit of data that can be processed by a Storm application is called a tuple.

Each tuple consists of a predefined list of fields. The value of each field can be a byte, char, integer, long, float, double, Boolean, or byte array. Storm also provides an API to define your own data types, which can be serialized as fields in a tuple.

A tuple is dynamically typed, that is, you just need to define the names of the fields

in a tuple and not their data type. The choice of dynamic typing helps to simplify

the API and makes it easy to use. Also, since a processing unit in Storm can process

multiple types of tuples, it's not practical to declare field types.

Each of the fields in a tuple can be accessed by its name getValueByField(String) or its positional index getValue(int) in the tuple. Tuples also provide convenient methods such as getIntegerByField(String) that save you from typecasting the objects. For example, if you have a Fraction(numerator, denominator) tuple, representing fractional numbers, then you can get the value of the numerator by either using getIntegerByField("numerator") or getInteger(0) .

You can see the full set of operations supported by backtype.storm.tuple.

backtype.storm.tuple.Tuple in the javadoc located at https://storm.

incubator.apache.org/apidocs/backtype/storm/tuple/Tuple.html .

Definition of a Storm topology

In Storm terminology, a topology is an abstraction that defines the graph of the computation. You create a Storm topology and deploy it on a Storm cluster to process the data. A topology can be represented by a direct acyclic graph, where each node does some kind of processing and forwards it to the next node(s) in the flow. The following is a sample Storm topology:

Graphical representation of the Storm topology

The following are the components of a Storm topology:

• Stream: The key abstraction in Storm is that of a stream. A stream is an unbounded sequence of tuples that can be processed in parallel by Storm.

Each stream can be processed by a single or multiple types of bolts (the

processing units in Storm, which are defined later in this section). Thus,

Storm can also be viewed as a platform to transform streams. In the

Each stream in a Storm application is given an ID and the bolts can produce and consume tuples from these streams on the basis of their ID. Each stream also has an associated schema for the tuples that will flow through it.

• Spout: A spout is the source of tuples in a Storm topology. It is responsible for reading or listening to data from an external source, for example, by reading from a logfile or listening for new messages in a queue and publishing them—emitting, in Storm terminology—into streams. A spout can emit multiple streams, each of different schemas. For example, it can read 10-field records from a logfile and emit them as different streams of 7-tuples and 4-tuples each.

The backtype.storm.spout.ISpout interface is the interface used to define spouts. If you are writing your topology in Java, then you should use backtype.storm.topology.IRichSpout as it declares methods to use the TopologyBuilder API. Whenever a spout emits a tuple, Storm tracks all the tuples generated while processing this tuple, and when the execution of all the tuples in the graph of this source tuple is complete, it will send back an acknowledgement to the spout. This tracking happens only if a message ID was provided while emitting the tuple. If null was used as message ID, this tracking will not happen.

A tuple-processing timeout can also be defined for a topology, and if a tuple is not processed within the specified timeout, a fail message will be sent back to the spout. Again, this will happen only if you define a message ID. A small performance gain can be extracted out of Storm at the risk of some data loss by disabling the message acknowledgements, which can be done by skipping the message ID while emitting tuples.

The important methods of spout are:

° nextTuple() : This method is called by Storm to get the next tuple

from the input source. Inside this method, you will have the logic

of reading data from the external sources and emitting them to an

instance of backtype.storm.spout.ISpoutOutputCollector .

The schema for streams can be declared by using the declareStream

method of backtype.storm.topology.OutputFieldsDeclarer .

If a spout wants to emit data to more than one stream, it can declare

multiple streams using the declareStream method and specify a

stream ID while emitting the tuple. If there are no more tuples to

emit at the moment, this method would not be blocked. Also, if this

method does not emit a tuple, then Storm will wait for 1 millisecond

before calling it again. This waiting time can be configured using the

topology.sleep.spout.wait.strategy.time.ms setting.

° ack(Object msgId) : This method is invoked by Storm when the tuple with the given message ID is completely processed by the topology. At this point, the user should mark the message as processed and do the required cleaning up such as removing the message from the message queue so that it does not get processed again.

° fail(Object msgId) : This method is invoked by Storm when it identifies that the tuple with the given message ID has not been processed successfully or has timed out of the configured interval.

In such scenarios, the user should do the required processing so that the messages can be emitted again by the nextTuple method.

A common way to do this is to put the message back in the incoming message queue.

° open() : This method is called only once—when the spout is initialized.

If it is required to connect to an external source for the input data, define the logic to connect to the external source in the open method, and then keep fetching the data from this external source in the nextTuple method to emit it further.

Another point to note while writing your spout is that none of the methods should be blocking, as Storm calls all the methods in the same thread. Every spout has an internal buffer to keep track of the status of the tuples emitted so far. The spout will keep the tuples in this buffer until they are either acknowledged or failed, calling the ack or fail method respectively.

Storm will call the nextTuple method only when this buffer is not full.

• Bolt: A bolt is the processing powerhouse of a Storm topology and is responsible for transforming a stream. Ideally, each bolt in the topology should be doing a simple transformation of the tuples, and many such bolts can coordinate with each other to exhibit a complex transformation.

The backtype.storm.task.IBolt interface is preferably used to define bolts, and if a topology is written in Java, you should use the backtype.

storm.topology.IRichBolt interface. A bolt can subscribe to multiple streams of other components—either spouts or other bolts—in the topology and similarly can emit output to multiple streams. Output streams can be declared using the declareStream method of backtype.storm.topology.

OutputFieldsDeclarer .

The important methods of a bolt are:

° execute(Tuple input) : This method is executed for each tuple that comes through the subscribed input streams. In this method, you can do whatever processing is required for the tuple and then produce the output either in the form of emitting more tuples to the declared output streams or other things such as persisting the results in a database.

You are not required to process the tuple as soon as this method is called, and the tuples can be held until required. For example, while joining two streams, when a tuple arrives, you can hold it until its counterpart also comes, and then you can emit the joined tuple.

The metadata associated with the tuple can be retrieved by the various methods defined in the Tuple interface. If a message ID is associated with a tuple, the execute method must publish an ack or fail event using OutputCollector for the bolt or else Storm will not know whether the tuple was processed successfully or not. The backtype.storm.topology.IBasicBolt interface is a convenient interface that sends an acknowledgement automatically after the completion of the execute method. In the case that a fail event is to be sent, this method should throw backtype.storm.topology.

FailedException .

° prepare(Map stormConf, TopologyContext context,

OutputCollector collector) : A bolt can be executed by multiple workers in a Storm topology. The instance of a bolt is created on the client machine and then serialized and submitted to Nimbus. When Nimbus creates the worker instances for the topology, it sends this serialized bolt to the workers. The work will desterilize the bolt and call the prepare method. In this method, you should make sure the bolt is properly configured to execute tuples now. Any state that you want to maintain can be stored as instance variables for the bolt that can be serialized/deserialized later.

Operation modes

Operation modes indicate how the topology is deployed in Storm. Storm supports two types of operation modes to execute the Storm topology

• The local mode: In the local mode, Storm topologies run on the local

machine in a single JVM. This mode simulates a Storm cluster in a single

JVM and is used for the testing and debugging of a topology.

• The remote mode: In the remote mode, we will use the Storm client to submit the topology to the master along with all the necessary code required to execute the topology. Nimbus will then take care of distributing your code.

Setting up your development environment

Before you can start developing Storm topologies, you must first check/set up your development environment, which involves installing the following software packages on your development computer:

• Java SDK 6

• Maven

• Git: Distributed version control

• Spring Tool Suite: IDE

The following installation steps are valid for CentOS, and going forward, all the commands used in this book are valid for CentOS.

Installing Java SDK 6

Perform the following steps to install the Java SDK 6 on your machine:

1. Download the Java SDK 6 RPM from Oracle's site ( http://www.oracle.

com/technetwork/java/javase/downloads/index.html ).

2. Install the Java jdk-6u31-linux-amd64.rpm file on your CentOS machine using the following command:

sudo rpm -ivh jdk-6u31-linux-amd64.rpm

3. Add the environment variable in the ~/.bashrc file:

export JAVA_HOME=/usr/java/jdk1.6.0_31/

4. Add the path of the bin directory of the JDK in the PATH system environment variable in the ~/.bashrc file:

export PATH=$JAVA_HOME/bin:$PATH

The PATH variable is the system variable that your operating system uses to locate the required executables from the command line or terminal window.

5. Run the following command to reload the bashrc file on the current

6. Check the Java installation as follows:

java -version

The output of the preceding command is:

java version "1.6.0_31"

Java(TM) SE Runtime Environment (build 1.6.0_31-b04)

Java HotSpot(TM) 64-Bit Server VM (build 20.6-b01, mixed mode)

Installing Maven

Apache Maven is a software dependency management tool and is used to manage the project's build, reporting, and documentation. We are using this so that we do not need to download all the dependencies manually. Perform the following steps to install the Maven on your machine:

1. Download the stable release of Maven from Maven's site ( http://maven.apache.org/download.cgi ).

2. Once you have downloaded the latest version, unzip it. Now, set the MAVEN_HOME environment variable in the ~/.bashrc file to make the setting up of Maven easier.

export MAVEN_HOME=/home/root/apache-maven-3.0.4

3. Add the path to the bin directory of Maven in the $PATH environment variable in the ~/.bashrc file:

export PATH=$JAVA_HOME/bin:$PATH:$MAVEN_HOME/bin

4. Run the following command to reload the bashrc file on the current login terminal:

source ~/.bashrc

5. Check the Maven installation as follows:

mvn –version

The following information will be displayed:

Apache Maven 3.0.4 (r1232337; 2012-01-17 14:14:56+0530) Maven home: /home/root/apache-maven-3.0.4

Java version: 1.6.0_31, vendor: Sun Microsystems Inc.

Java home: /usr/java/jdk1.6.0_31/jre

Default locale: en_US, platform encoding: UTF-8

OS name: "linux", version: "2.6.32-279.22.1.el6.x86_64", arch:

"amd64", family: "unix"

Installing Git – distributed version control

Git is one of the most used open source version control systems. It is used to track content such as files and directories and allows multiple users to work on the same file. Perform the following steps to install Git on your machine:

1. The command to install Git on a CentOS machine is:

sudo yum install git

2. Check the installation of Git using the following command:

git --version

The preceding command's output is:

git version 1.7.1

Installing the STS IDE

The STS IDE is an integrated development environment and is used to develop applications. We will be using this to develop all the examples in this book.

Perform the following steps to install the STS IDE on your machine:

1. Download the latest version of STS from the Spring site ( https://spring.io/tools/sts/all ).

2. Once you have downloaded the latest version, unzip it.

3. Start the STS IDE.

4. Go to Windows | Preferences | Maven | Installations and add the path of maven-3.0.4 , as shown in the following screenshot:

Add maven-3.0.4 to launch Maven

5. Go to Window | Preferences | Java | Installed JREs and add the path of Java Runtime Environment 6 (JRE 6), as shown in the following screenshot:

Add jdk1.6.0_31 to the build path

From now on, we will use the Spring Tool Suite to develop all the sample Storm topologies.

Developing a sample topology

The sample topology shown in the following diagram will cover how to create a basic Storm project, including a spout and bolt, build it, and execute it:

Spout Bolt

Hello World Topology

This spout continuously emits strings as tuples chosen randomly from a

map.

This bolt read the stream subscribed by a bolt and print the tuples value.

A sample Hello World topology

Perform the following steps to create and execute a sample topology:

1. Start your STS IDE and create a Maven project as shown in the following screenshot:

Create a Maven project

2. Specify com.learningstorm as Group Id and storm-example as Artifact Id, as shown in the following screenshot:

Specify Archetype Parameters

3. Add the following Maven dependencies in the pom.xml file:

<dependencies>

<dependency>

<groupId>junit</groupId>

<artifactId>junit</artifactId>

<version>3.8.1</version>

<scope>test</scope>

</dependency>

<dependency>

<groupId>storm</groupId>

<artifactId>storm</artifactId>

<version>0.9.0.1</version>

<scope>provided</scope>

</dependency>

</dependencies>

4. Add the following Maven repository in the pom.xml file:

<repositories>

<repository>

<id>clojars.org</id>

<url>http://clojars.org/repo</url>

</repository>

</repositories>

5. Add the following Maven build plugins in the pom.xml file:

<build>

<plugins>

<plugin>

<artifactId>maven-assembly-plugin</artifactId>

<version>2.2.1</version>

<configuration>

<descriptorRefs>

<descriptorRef>jar-with-dependencies </descriptorRef>

</descriptorRefs>

<archive>

<manifest>

<mainClass />

</manifest>

</archive>

</configuration>

<executions>

<execution>

<goals>

<goal>single</goal>

</goals>

</execution>

</executions>

</plugin>

</plugins>

</build>

6. Write your first sample spout by creating a LearningStormSpout class in the com.learningstorm.storm_example package. The LearningStormSpout class extends the serialized BaseRichSpout class. This spout does not connect to an external source to fetch data but randomly generates the data and emits a continuous stream of records. The following is the source code of the LearningStormSpout class with an explanation:

public class LearningStormSpout extends BaseRichSpout{

private static final long serialVersionUID = 1L;

private SpoutOutputCollectorspoutOutputCollector;

private static final Map<Integer, String> map = new HashMap<Integer, String>();

static {

map.put(0, "google");

map.put(1, "facebook");

map.put(2, "twitter");

map.put(3, "youtube");

map.put(4, "linkedin");

}

public void open(Map conf, TopologyContext context, SpoutOutputCollector spoutOutputCollector) {

// Open the spout

this.spoutOutputCollector = spoutOutputCollector;

}

public void nextTuple() {

// Storm cluster repeatedly calls this method to emit a continuous

// stream of tuples.

final Random rand = new Random();

// generate the random number from 0 to 4.

int randomNumber = rand.nextInt(5);

spoutOutputCollector.emit(new Values(map.get(randomNumber)));

}

public void declareOutputFields(OutputFieldsDeclarer

declarer) {

// emit the tuple with field "site"

declarer.declare(new Fields("site"));

} }

7. Write your first sample bolt by creating a LearningStormBolt class within the same package. The LearningStormBolt class extends the serialized BaseRichBolt class. This bolt will consume the tuples emitted by LearningStormSpout spout and will print the value of the field "site"

on the console. The following is the source code of the LearningStormBolt class with an explanation:

public class LearningStormBolt extends BaseBasicBolt{

private static final long serialVersionUID = 1L;

public void execute(Tuple input, BasicOutputCollector collector) {

// fetched the field "site" from input tuple.

String test = input.getStringByField("site");

// print the value of field "site" on console.

System.out.println("Name of input site is : " + test);

}

public void declareOutputFields(OutputFieldsDeclarer declarer) {

} }

8. Create a main LearningStormTopology class within the same package. This class creates an instance of the spout and bolt, classes and chained together using a TopologyBuilder class. The following is the implementation of the main class:

public class LearningStormTopology {

public static void main(String[] args) throws AlreadyAliveException, InvalidTopologyException { // create an instance of TopologyBuilder class TopologyBuilder builder = new TopologyBuilder();

// set the spout class

builder.setSpout("LearningStormSpout", new LearningStormSpout(), 2);

// set the bolt class

builder.setBolt("LearningStormBolt",

new LearningStormBolt(), 4).shuffleGrouping

Config conf = new Config();

conf.setDebug(true);

// create an instance of LocalCluster class for // executing topology in local mode.

LocalCluster cluster = new LocalCluster();

// LearningStormTopolgy is the name of submitted topology.

cluster.submitTopology("LearningStormToplogy", conf, builder.createTopology());

try {

Thread.sleep(10000);

} catch (Exception exception) {

System.out.println("Thread interrupted exception : "

+ exception);

}

// kill the LearningStormTopology

cluster.killTopology("LearningStormToplogy");

// shutdown the storm test cluster cluster.shutdown();

} }

9. Go to your project's home directory and run the following commands to execute the topology in the local mode:

mvn compile exec:java -Dexec.classpathScope=compile -

Dexec.mainClass=com.learningstorm.storm_example.

LearningStormTopology

Downloading the example code

You can download the example code files for all Packt books you have purchased from your account at http://www.packtpub.com.

If you purchased this book elsewhere, you can visit http://www.

packtpub.com/support and register to have the files e-mailed directly to you.

Also, we can execute the topology by simply running the main class through the STS IDE.

In the preceding example, we used a utility called LocalCluster to execute the

topology in a single JVM. The LocalCluster class simulates the Storm cluster

and starts all the Storm processes in a single JVM.

We have submitted a topology in a simulated cluster by calling the submitTopology method of the LocalCluster class. The submitTopology method takes the name of a topology, a configuration for the topology, and then the topology itself as arguments.

The topology name is used to identify the topology in the Storm cluster. Hence, it is good practice to use a unique name for each topology.

Running the Storm infrastructure in local mode is useful when we want to test and debug the topology.

The upcoming sections will cover the deployment of ZooKeeper, Storm native dependencies, and Storm, and how we can submit the topology on a single-node Storm cluster.

Setting up ZooKeeper

This section describes how you can set up a ZooKeeper cluster. We are deploying ZooKeeper in standalone mode, but in the distributed cluster mode, it is always recommended that you should run a ZooKeeper ensemble of at least three nodes to support failover and high availability. Perform the following steps to set up ZooKeeper on your machine:

1. Download the latest stable ZooKeeper release from the ZooKeeper's site ( http://www.apache.org/dyn/closer.cgi/zookeeper/ ); at this moment, the latest version is ZooKeeper 3.4.5.

2. Once you have downloaded the latest version, unzip it and set the ZK_HOME environment variable.

3. Create the configuration file, zoo.cfg , at the $ZK_HOME/conf directory using the following command:

cd $ZK_HOME/conf touch zoo.cfg

4. Add the following three properties in the zoo.cfg file:

tickTime=2000

dataDir=/tmp/zookeeper clientPort=2181

The following are the definitions of each of these properties:

° tickTime : This is the basic time unit in milliseconds used by

ZooKeeper. It is used to send heartbeats and the minimum session

timeout will be twice the value.

° dataDir : This is an empty directory to store the in-memory database snapshots and transactional log.

° clientPort : This is the port used to listen for client connections.

5. The command to start the ZooKeeper node is as follows:

bin/zkServer.sh start

The following information is displayed:

JMX enabled by default

Using config: /home/root/zookeeper-3.4.5/bin/../conf/zoo.cfg Starting zookeeper ... STARTED

6. At this point, the following Java process must be started:

jps

The following information is displayed:

23074 QuorumPeerMain

7. The command to check the status of running the ZooKeeper node is as follows:

bin/zkServer.sh status

The following information is displayed:

JMX enabled by default

Using config: ../conf/zoo.cfg Mode: standalone

Setting up Storm on a single development machine

This section describes you how to install Storm on a single machine. Download the latest stable Storm release from https://storm.incubator.apache.org/

downloads.html ; at the time of this writing, the latest version is storm-0.9.0.1.

Perform the following steps to set up Storm on a single development machine:

1. Once you have downloaded the latest version, unzip it and set the STORM_HOME environment variable.

2. Perform the following steps to edit the storm.yaml configuration file:

cd $STORM_HOME/conf

vi storm.yaml

Add the following information:

storm.zookeeper.servers:

- "127.0.0.1"

storm.zookeeper.port: 2181 nimbus.host: "127.0.0.1"

storm.local.dir: "/tmp/storm-data"

java.library.path: "/usr/local/lib"

storm.messaging.transport: backtype.storm.messaging.netty.Context supervisor.slots.ports:

- 6700 - 6701 - 6702 - 6703

3. The following is a definition of properties used in the storm.yaml file:

° storm.zookeeper.servers : This property contains the IP addresses of ZooKeeper servers.

° storm.zookeeper.port : This property contains the ZooKeeper client port.

° storm.local.dir : The Nimbus and supervisor daemons require a directory on the local disk to store small amounts of state (such as JARs, CONFs, and more).

° java.library.path : This is used to load the Java native libraries that Storm uses (ZeroMQ and JZMQ). The default location of Storm native libraries is /usr/local/lib : /opt/local/lib:/usr/lib . ° nimbus.host : This specifies the IP address of the master

(Nimbus) node:

° supervisor.slots.ports : For every worker machine, we can configure how many workers run on that machine with this property.

Each worker binds with a single port and uses that port to receive incoming messages.

4. Start the master node using the following commands:

cd $STORM_HOME bin/storm nimbus

5. Start the supervisor node using the following commands:

bin/storm supervisor

Deploying the sample topology on a single-node cluster

In the previous example, we executed the Storm topology in the local mode.

Now, we will deploy the topology on the single-node Storm cluster.

1. We will first create a LearningStormSingleNodeTopology class within the same package. The following LearningStormSingleNodeTopology class will use the submitTopology method of the StormSubmitter class to deploy the topology on the Storm cluster:

public class LearningStormSingleNodeTopology { public static void main(String[] args) {

TopologyBuilder builder = new TopologyBuilder();

// set the spout class

builder.setSpout("LearningStormSpout", new LearningStormSpout(), 4);

// set the bolt class

builder.setBolt("LearningStormBolt", new LearningStormBolt(), 2)

.shuffleGrouping("LearningStormSpout");

Config conf = new Config();

conf.setNumWorkers(3);

try {

// This statement submit the topology on remote cluster.

// args[0] = name of topology

StormSubmitter.submitTopology(args[0], conf, builder.createTopology());

}catch(AlreadyAliveException alreadyAliveException) { System.out.println(alreadyAliveException);

} catch

(InvalidTopologyException invalidTopologyException) { System.out.println(invalidTopologyException);

} } }

2. Build your Maven project by running the following command on the project home directory:

mvn clean install

The output of the preceding command is:

--- ---

[INFO] --- ---

[INFO] BUILD SUCCESS

[INFO] --- ---

[INFO] Total time: 58.326s

[INFO] Finished at: Mon Jan 20 00:55:52 IST 2014 [INFO] Final Memory: 14M/116M

[INFO] --- ---

3. We can deploy the topology to the cluster using the following Storm client command:

bin/storm jar jarName.jar [TopologyMainClass] [Args]

The preceding command runs TopologyMainClass with the arguments, arg1 and arg2 . The main function of TopologyMainClass is to define the topology and submit it to Nimbus. The Storm JAR part takes care of connecting to Nimbus and uploading the JAR part.

4. Go to the $STORM_HOME directory and run the following command to deploy LearningStormSingleNodeTopology to the Storm cluster:

bin/storm jar $PROJECT_HOME/target/storm-example-0.0.1-SNAPSHOT- jar-with-dependencies.jar com.learningstorm.storm_example.

LearningStormSingleNodeTopology LearningStormSingleNodeTopology The following information is displayed:

0 [main] INFO backtype.storm.StormSubmitter - Jar not uploaded to master yet. Submitting jar...

7 [main] INFO backtype.storm.StormSubmitter - Uploading topology jar /home/root/storm-example/target/storm-example- 0.0.1-SNAPSHOT-jar-with-dependencies.jar to assigned location: / tmp/storm-data/nimbus/inbox/stormjar-dfce742b-ca0b-4121-bcbe- 1856dc1846a4.jar

19 [main] INFO backtype.storm.StormSubmitter - Successfully uploaded topology jar to assigned location: /tmp/storm-data/

nimbus/inbox/stormjar-dfce742b-ca0b-4121-bcbe-1856dc1846a4.jar

19 [main] INFO backtype.storm.StormSubmitter - Submitting

219 [main] INFO backtype.storm.StormSubmitter - Finished submitting topology: LearningStormSingleNodeTopology

5. Run the jps command to see the number of running JVM processes as follows:

jps

The preceding command's output is:

26827 worker 26530 supervisor 26824 worker 26468 nimbus

15987 QuorumPeerMain 26822 worker

6. Storm supports deactivating a topology. In the deactivated state, spouts will not emits any new tuples into pipeline, but the processing of already emitted tuples will continue. The following is the command to deactivate the running topology:

bin/storm deactivate topologyName

7. Deactivate LearningStormSingleNodeTopology using the following command:

bin/storm deactivate LearningStormSingleNodeTopology The following information is displayed:

0 [main] INFO backtype.storm.thrift - Connecting to Nimbus at localhost:6627r

76 [main] INFO backtype.storm.command.deactivate - Deactivated topology: LearningStormSingleNodeTopology

8. Storm also supports activating a topology. When a topology is activated, spouts will again start emitting tuples. The following is the command to activate the topology:

bin/storm activate topologyName

9. Activate LearningStormSingleNodeTopology using the following command:

bin/storm activate LearningStormSingleNodeTopology The following information is displayed:

0 [main] INFO backtype.storm.thrift - Connecting to Nimbus at localhost:6627

65 [main] INFO backtype.storm.command.activate - Activated

topology: LearningStormSingleNodeTopology

10. Storm topologies are never-ending processes. To stop a topology, we need to kill it. When killed, the topology first enters into the deactivation state, processes all the tuples already emitted into it, and then stops. Run the following command to kill LearningStormSingleNodeTopology : bin/storm kill LearningStormSingleNodeTopology

The following information is displayed:

0 [main] INFO backtype.storm.thrift - Connecting to Nimbus at localhost:6627

80 [main] INFO backtype.storm.command.kill-topology - Killed topology: LearningStormSingleNodeTopology

11. Now, run the jps command again to see the remaining JVM processes as follows:

jps

The preceding command's output is:

26530 supervisor 27193 Jps

26468 nimbus

15987 QuorumPeerMain

12. To update a running topology, the only option available is to kill the currently running topology and submit a new one.

Summary

In this chapter, we introduced you to the basics of Storm and the various components that make up a Storm cluster. We saw the different operation modes in which a Storm cluster can operate. We deployed a single-node Storm cluster and also developed a sample topology to run it on the single-node Storm cluster.

In the next chapter, we will set up a three-node Storm cluster to run the sample

topology. We will also see different types of Stream groupings supported by

Storm and the guaranteed message semantic provided by Storm.

Setting Up a Storm Cluster

In the last chapter, we saw how to write a minimal Storm topology and run it on the local mode and a single-node Storm cluster. In this chapter, we will cover the following topics:

• How to run the sample topology in a distributed Storm cluster

• How to configure the parallelism of a topology

• How to partition a stream using different stream grouping

In the last chapter, we saw how to set up single-node ZooKeeper to use with Storm.

Even though we can proceed with the same ZooKeeper setup for a distributed Storm cluster setup, then it will be a single point of failure in the cluster. To avoid this, we are deploying a distributed ZooKeeper cluster.

It is advised to run an odd number of ZooKeeper nodes, as the ZooKeeper cluster

keeps working as long as the majority (the number of live nodes is greater than n/2,

where n is the number of deployed nodes) of the nodes are running. So, if we have a

cluster of four ZooKeeper nodes (3 > 4/2, only one node can die), then we can handle

only one node failure, while if we had five nodes (3 > 5/2, two nodes can die) in the

cluster, we can handle two node failures.

We will be deploying a ZooKeeper ensemble of three nodes that will handle one node failure. The following is the deployment diagram of the three-node ZooKeeper ensemble:

Follower Leader Follower

Server1

(zoo1) Server2

(zoo2) Server3

(zoo3)

client client client client

Zoo eeper Ensemble K

A ZooKeeper ensemble

In the ZooKeeper ensemble, one node in the cluster acts as the leader, while the rest are followers. If the leader node of the ZooKeeper cluster dies, then an election for the new leader takes places among the remaining live nodes, and a new leader is elected. All write requests coming from clients are forwarded to the leader node, while the follower nodes only handle the read requests. Also, we can't increase the write performance of the ZooKeeper ensemble by increasing the number of nodes because all write operations go through the leader node.

The following steps need to be performed on each node to deploy the ZooKeeper ensemble:

1. Download the latest stable ZooKeeper release from the ZooKeeper site ( http://zookeeper.apache.org/releases.html ). At this moment, the latest version is ZooKeeper 3.4.5.

2. Once you have downloaded the latest version, unzip it. Now, we set up the ZK_HOME environment variable to make the setup easier.

3. Point the ZK_HOME environment variable to the unzipped directory. Create the configuration file, zoo.cfg , at $ZK_HOME/conf directory using the following commands:

cd $ZK_HOME/conf

touch zoo.cfg