Apache Kafka
Set up Apache Kafka clusters and develop custom message producers and consumers using practical, hands-on examples
Nishant Garg
BIRMINGHAM - MUMBAI
Copyright © 2013 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 author, 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: October 2013
Production Reference: 1101013
Published by Packt Publishing Ltd.
Livery Place 35 Livery Street
Birmingham B3 2PB, UK..
ISBN 978-1-78216-793-8 www.packtpub.com
Cover Image by Suresh Mogre ( [email protected] )
Credits
Author Nishant Garg
Reviewers Magnus Edenhill Iuliia Proskurnia
Acquisition Editors Usha Iyer
Julian Ursell
Commissioning Editor Shaon Basu
Technical Editor Veena Pagare
Copy Editors Tanvi Gaitonde Sayanee Mukherjee Aditya Nair
Kirti Pai Alfida Paiva Adithi Shetty
Project Coordinator Esha Thakker
Proofreader Christopher Smith
Indexers Monica Ajmera Hemangini Bari Tejal Daruwale
Graphics Abhinash Sahu
Production Coordinator Kirtee Shingan
Cover Work
Kirtee Shingan
About the Author
Nishant Garg is a Technical Architect with more than 13 years' experience in various technologies such as Java Enterprise Edition, Spring, Hibernate, Hadoop, Hive, Flume, Sqoop, Oozie, Spark, Kafka, Storm, Mahout, and Solr/Lucene; NoSQL databases such as MongoDB, CouchDB, HBase and Cassandra, and MPP Databases such as GreenPlum and Vertica.
He has attained his M.S. in Software Systems from Birla Institute of Technology and Science, Pilani, India, and is currently a part of Big Data R&D team in innovation labs at Impetus Infotech Pvt. Ltd.
Nishant has enjoyed working with recognizable names in IT services and financial industries, employing full software lifecycle methodologies such as Agile and SCRUM.
He has also undertaken many speaking engagements on Big Data technologies.
I would like to thank my parents (Sh. Vishnu Murti Garg and Smt.
Vimla Garg) for their continuous encouragement and motivation throughout my life. I would also like to thank my wife (Himani) and my kids (Nitigya and Darsh) for their never-ending support, which keeps me going.
Finally, I would like to thank Vineet Tyagi—AVP and Head of
Innovation Labs, Impetus—and Dr. Vijay—Director of Technology,
Innovation Labs, Impetus—for having faith in me and giving me
an opportunity to write.
About the Reviewers
Magnus Edenhill is a freelance systems developer living in Stockholm, Sweden, with his family. He specializes in high-performance distributed systems but is also a veteran in embedded systems.
For ten years, Magnus played an instrumental role in the design and implementation of PacketFront's broadband architecture, serving millions of FTTH end customers worldwide. Since 2010, he has been running his own consultancy business with customers ranging from Headweb—northern Europe's largest movie streaming service—to Wikipedia.
Iuliia Proskurnia is a doctoral student at EDIC school of EPFL, specializing in Distributed Computing. Iuliia was awarded the EPFL fellowship to conduct her doctoral research. She is a winner of the Google Anita Borg scholarship and was the Google Ambassador at KTH (2012-2013). She obtained a Masters Diploma in Distributed Computing (2013) from KTH, Stockholm, Sweden, and UPC,
Barcelona, Spain. For her Master's thesis, she designed and implemented a unique
real-time, low-latency, reliable, and strongly consistent distributed data store
for the stock exchange environment at NASDAQ OMX. Previously, she has
obtained Master's and Bachelor's Diplomas with honors in Computer Science
from the National Technical University of Ukraine KPI. This Master's thesis was
about fuzzy portfolio management in previously uncertain conditions. This period
was productive for her in terms of publications and conference presentations. During
her studies in Ukraine, she obtained several scholarships. During her stay in Kiev,
Ukraine, she worked as Financial Analyst at Alfa Bank Ukraine.
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: Introducing Kafka 5
Need for Kafka 7
Few Kafka usages 8
Summary 9
Chapter 2: Installing Kafka 11
Installing Kafka 12
Downloading Kafka 12
Installing the prerequisites 13
Installing Java 1.6 or later 13
Building Kafka 14
Summary 16 Chapter 3: Setting up the Kafka Cluster 17 Single node – single broker cluster 17
Starting the ZooKeeper server 18
Starting the Kafka broker 19
Creating a Kafka topic 20
Starting a producer for sending messages 20
Starting a consumer for consuming messages 22
Single node – multiple broker cluster 23
Starting ZooKeeper 23
Starting the Kafka broker 23
Creating a Kafka topic 24
Starting a producer for sending messages 24
Starting a consumer for consuming messages 25
Multiple node – multiple broker cluster 25
Kafka broker property list 26
Summary 26
Chapter 4: Kafka Design 27
Kafka design fundamentals 28
Message compression in Kafka 29 Cluster mirroring in Kafka 30
Replication in Kafka 31
Summary 32 Chapter 5: Writing Producers 33
The Java producer API 34
Simple Java producer 36
Importing classes 36
Defining properties 36
Building the message and sending it 37
Creating a simple Java producer with message partitioning 38
Importing classes 38
Defining properties 38
Implementing the Partitioner class 39
Building the message and sending it 39
The Kafka producer property list 40 Summary 42 Chapter 6: Writing Consumers 43
Java consumer API 44
High-level consumer API 44
Simple consumer API 46
Simple high-level Java consumer 47
Importing classes 47
Defining properties 47
Reading messages from a topic and printing them 48 Multithreaded consumer for multipartition topics 50
Importing classes 50
Defining properties 50
Reading the message from threads and printing it 51 Kafka consumer property list 54 Summary 55 Chapter 7: Kafka Integrations 57 Kafka integration with Storm 57
Introduction to Storm 58
Integrating Storm 59
Kafka integration with Hadoop 60
Introduction to Hadoop 60
Integrating Hadoop 62
Table of Contents
[ iii ]
Hadoop producer 62
Hadoop consumer 64
Summary 64
Chapter 8: Kafka Tools 65
Kafka administration tools 65
Kafka topic tools 65
Kafka replication tools 66
Integration with other tools 68
Kafka performance testing 69
Summary 69
Index 71
Preface
This book is here to help you get familiar with Apache Kafka and use it to solve your challenges related to the consumption of millions of messages in publisher-subscriber architecture. It is aimed at getting you started with a feel for programming with Kafka so that you will have a solid foundation to dive deep into its different types
of implementations and integrations.
In addition to an explanation of Apache Kafka, we also offer a chapter exploring Kafka integration with other technologies such as Apache Hadoop and Storm. Our goal is to give you an understanding of not just what Apache Kafka is, but also how to use it as part of your broader technical infrastructure.
What this book covers
Chapter 1, Introducing Kafka, discusses how organizations are realizing the real value of data and evolving the mechanism of collecting and processing it.
Chapter 2, Installing Kafka, describes how to install and build Kafka 0.7.x and 0.8.
Chapter 3, Setting up the Kafka Cluster, describes the steps required to set up a single/multibroker Kafka cluster.
Chapter 4, Kafka Design, discusses the design concepts used for building a solid foundation for Kafka.
Chapter 5, Writing Producers, provides detailed information about how to write basic producers and some advanced-level Java producers that use message partitioning.
Chapter 6, Writing Consumers, provides detailed information about how to write basic
consumers and some advanced-level Java consumers that consume messages from
the partitions.
Chapter 7, Kafka Integrations, discusses how Kafka integration works for both Storm and Hadoop to address real-time and batch processing needs.
Chapter 8, Kafka Tools, describes information about Kafka tools, such as its
administrator tools, and Kafka integration with Camus, Apache Camel, Amazon cloud, and so on.
What you need for this book
In the simplest case, a single Linux-based (CentOS 6.x) machine with JDK 1.6 installed will give you a platform to explore almost all the exercises in this book.
We assume you have some familiarity with command-line Linux; any modern distribution will suffice.
Some of the examples in this book need multiple machines to see things working, so you will require access to at least three such hosts. Virtual machines are fine for learning and exploration.
You will generally need the big data technologies, such as Hadoop and Storm, to run your Hadoop and Storm clusters.
Who this book is for
This book is for readers who want to know about Apache Kafka at a hands-on level; the key audience is those with software development experience but no prior exposure to Apache Kafka or similar technologies.
This book is also for enterprise application developers and big data enthusiasts who have worked with other publisher-subscriber-based systems and now want to explore Apache Kafka as a futuristic scalable solution.
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 are shown as follows: "We can include other contexts through
the use of the include directive."
Preface
[ 3 ] A block of code is set as follows:
String messageStr = new String("Hello from Java Producer");
KeyedMessage<Integer, String> data = new KeyedMessage<Integer, String>(topic, messageStr);
producer.send(data);
When we wish to draw your attention to a particular part of a code block, the relevant lines or items are set in bold:
Properties props = new Properties();
props.put("metadata.broker.list","localhost:9092");
props.put("serializer.class","kafka.serializer.StringEncoder");
props.put("request.required.acks", "1");
ProducerConfig config = new ProducerConfig(props);
Producer<Integer, String> producer = new Producer<Integer, String>(config);
Any command-line input or output is written as follows:
[root@localhost kafka-0.8]# java SimpleProducer kafkatopic Hello_There
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 color images of this book
We also provide you with a PDF file that has color images of the screenshots used in this book. You can download this file from http://www.packtpub.com/sites/
default/files/downloads/7938OS_Images.pdf .
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.
Introducing Kafka
Welcome to the world of Apache Kafka.
In today's world, real-time information is continuously getting generated by applications (business, social, or any other type), and this information needs easy ways to be reliably and quickly routed to multiple types of receivers. Most of the time, applications that are producing information and applications that are consuming this information are well apart and inaccessible to each other. This, at times, leads to redevelopment of information producers or consumers to provide an integration point between them. Therefore, a mechanism is required for seamless integration of information of producers and consumers to avoid any kind of
rewriting of an application at either end.
In the present big data era, the very first challenge is to collect the data as it is a huge amount of data and the second challenge is to analyze it. This analysis typically includes following type of data and much more:
• User behavior data
• Application performance tracing
• Activity data in the form of logs
• Event messages
Message publishing is a mechanism for connecting various applications with the help of messages that are routed between them, for example, by a message broker such as Kafka. Kafka is a solution to the real-time problems of any software solution, that is, to deal with real-time volumes of information and route it to multiple
consumers quickly. Kafka provides seamless integration between information
of producers and consumers without blocking the producers of the information,
and without letting producers know who the final consumers are.
Apache Kafka is an open source, distributed publish-subscribe messaging system, mainly designed with the following characteristics:
• Persistent messaging: To derive the real value from big data, any kind of information loss cannot be afforded. Apache Kafka is designed with O(1) disk structures that provide constant-time performance even with very large volumes of stored messages, which is in order of TB.
• High throughput: Keeping big data in mind, Kafka is designed to work on commodity hardware and to support millions of messages per second.
• Distributed: Apache Kafka explicitly supports messages partitioning over Kafka servers and distributing consumption over a cluster of consumer machines while maintaining per-partition ordering semantics.
• Multiple client support: Apache Kafka system supports easy integration of clients from different platforms such as Java, .NET, PHP, Ruby, and Python.
• Real time: Messages produced by the producer threads should be immediately visible to consumer threads; this feature is critical to event-based systems such as Complex Event Processing (CEP) systems.
Kafka provides a real-time publish-subscribe solution, which overcomes the challenges of real-time data usage for consumption, for data volumes that may grow in order of magnitude, larger that the real data. Kafka also supports parallel data loading in the Hadoop systems.
The following diagram shows a typical big data aggregation-and-analysis scenario supported by the Apache Kafka messaging system:
Producer
(Front End) Producer
(Services) Producer
(Proxies) Producer
(Adapters) Other Producer
Consumers
(Real Time) Consumers
(NoSQL) Consumers
(Hadoop) Consumers
(Warehouses) Other Producer
ZooKeeper
Kafka Broker Kafka Broker Kafka Broker Kafka Broker
Chapter 1
[ 7 ]
At the production side, there are different kinds of producers, such as the following:
• Frontend web applications generating application logs
• Producer proxies generating web analytics logs
• Producer adapters generating transformation logs
• Producer services generating invocation trace logs
At the consumption side, there are different kinds of consumers, such as the following:
• Offline consumers that are consuming messages and storing them in Hadoop or traditional data warehouse for offline analysis
• Near real-time consumers that are consuming messages and storing them in any NoSQL datastore such as HBase or Cassandra for near real-time analytics
• Real-time consumers that filter messages in the in-memory database and trigger alert events for related groups
Need for Kafka
A large amount of data is generated by companies having any form of web-based presence and activity. Data is one of the newer ingredients in these Internet-based systems and typically includes user-activity events corresponding to logins, page visits, clicks, social networking activities such as likes, sharing, and comments, and operational and system metrics. This data is typically handled by logging and traditional log aggregation solutions due to high throughput (millions of messages per second). These traditional solutions are the viable solutions for providing logging data to an offline analysis system such as Hadoop. However, the solutions are very limiting for building real-time processing systems.
According to the new trends in Internet applications, activity data has become a part of production data and is used to run analytics at real time. These analytics can be:
• Search based on relevance
• Recommendations based on popularity, co-occurrence, or sentimental analysis
• Delivering advertisements to the masses
• Internet application security from spam or unauthorized data scraping
Real-time usage of these multiple sets of data collected from production systems has become a challenge because of the volume of data collected and processed.
Apache Kafka aims to unify offline and online processing by providing a mechanism for parallel load in Hadoop systems as well as the ability to partition real-time
consumption over a cluster of machines. Kafka can be compared with Scribe or Flume as it is useful for processing activity stream data; but from the architecture perspective, it is closer to traditional messaging systems such as ActiveMQ or RabitMQ.
Few Kafka usages
Some of the companies that are using Apache Kafka in their respective use cases are as follows:
• LinkedIn ( www.linkedin.com ): Apache Kafka is used at LinkedIn for the streaming of activity data and operational metrics. This data powers various products such as LinkedIn news feed and LinkedIn Today in addition to offline analytics systems such as Hadoop.
• DataSift ( www.datasift.com/ ): At DataSift, Kafka is used as a collector for monitoring events and as a tracker of users' consumption of data streams in real time.
• Twitter ( www.twitter.com/ ): Twitter uses Kafka as a part of its Storm— a stream-processing infrastructure.
• Foursquare ( www.foursquare.com/ ): Kafka powers online-to-online and online-to-offline messaging at Foursquare. It is used to integrate Foursquare monitoring and production systems with Foursquare, Hadoop-based offline infrastructures.
• Square ( www.squareup.com/ ): Square uses Kafka as a bus to move all system events through Square's various datacenters. This includes metrics, logs, custom events, and so on. On the consumer side, it outputs into Splunk, Graphite, or Esper-like real-time alerting.
The source of the above information is https://cwiki.
apache.org/confluence/display/KAFKA/Powered+By.
Chapter 1
[ 9 ]
Summary
In this chapter, we have seen how companies are evolving the mechanism of collecting and processing application-generated data, and that of utilizing the real power of this data by running analytics over it.
In the next chapter we will look at the steps required to install Kafka.
Installing Kafka
Kafka is an Apache project and its current Version 0.7.2 is available as a stable release. Kafka Version 0.8 is available as beta release, which is gaining acceptance in many large-scale enterprises. Kafka 0.8 offers many advanced features compared to 0.7.2. A few of its advancements are as follows:
• Prior to 0.8, any unconsumed partition of data within the topic could be lost if the broker failed. Now the partitions are provided with a replication factor.
This ensures that any committed message would not be lost, as at least one replica is available.
• The previous feature also ensures that all the producers and consumers are replication aware. By default, the producer's message send request is blocked until the message is committed to all active replicas; however, producers can also be configured to commit messages to a single broker.
• Like Kafka producers, Kafka consumers' polling model changes to a long pulling model and gets blocked until a committed message is available from the producer, which avoids frequent pulling.
• Additionally, Kafka 0.8 also comes with a set of administrative tools, such
as controlled shutdown of cluster and Lead replica election tool, for
managing the Kafka cluster.
The major limitation is that Kafka Version 0.7.x can't just be replaced by Version 0.8, as it is not backward compatible. If the existing Kafka cluster is based on 0.7.x, a migration tool is provided for migrating the data from the Kafka 0.7.x-based cluster to the 0.8-based cluster. This migration tool actually works as a consumer for 0.7.x-based Kafka clusters and republishes the messages as a producer to Kafka 0.8-based clusters. The following diagram explains this migration:
Kafka 0.7.x Cluster
Kafka 0.7.x Consumer
Kafka Migration
Kafka 0.8 Producer
Kafka 0.8 Cluster
More information about Kafka migration from 0.7.x to 0.8 can be found at https://cwiki.apache.org/confluence/display/KAFKA/
Migrating+from+0.7+to+0.8.
Coming back to installing Kafka, as a first step, we need to download the available stable/beta release (all the commands are tested on CentOS 5.5 OS and may differ on other kernel-based OS).
Installing Kafka
Now let us see what steps need to be followed in order to install Kafka:
Downloading Kafka
Perform the following steps for downloading Kafka release 0.7.x:
1. Download the current stable version of Kafka (0.7.2) into a folder on your file system (for example, /opt ) using the following command:
[root@localhost opt]#wget https://www.apache.org/dyn/closer.cgi/
incubator/kafka/kafka-0.7.2-incubating/kafka-0.7.2-incubating-src.
tgz
2. Extract the downloaded kafka-0.7.2-incubating-src.tgz using the following command:
[root@localhost opt]# tar xzf kafka-0.7.2-incubating-src.tgz
Chapter 2
[ 13 ]
Perform the following steps for downloading Kafka release 0.8:
1. Download the current beta release of Kafka (0.8) into a folder on your filesystem (for example, /opt ) using the following command:
[root@localhost opt]#wget
https://dist.apache.org/repos/dist/release/kafka/kafka-0.8.0- beta1-src.tgz
2. Extract the downloaded kafka-0.8.0-beta1-src.tgz using the following command:
[root@localhost opt]# tar xzf kafka-0.8.0-beta1-src.tgz
Going forward, all commands in this chapter are same for both the versions (0.7.x and 0.8) of Kafka.
Installing the prerequisites
Kafka is implemented in Scala and uses the ./sbt tool for building Kafka binaries.
sbt is a build tool for Scala and Java projects which requires Java 1.6 or later.
Installing Java 1.6 or later
Perform the following steps for installing Java 1.6 or later:
1. Download the jdk-6u45-linux-x64.bin link from Oracle's website:
http://www.oracle.com/technetwork/java/javase/downloads/index.
html .
2. Make sure the file is executable:
[root@localhost opt]#chmod +x jdk-6u45-linux-x64.bin 3. Run the installer:
[root@localhost opt]#./jdk-6u45-linux-x64.bin
4. Finally, add the environment variable JAVA_HOME . The following command will write the JAVA_HOME environment variable to the file /etc/profile , which contains system-wide environment configuration:
[root@localhost opt]# echo "export JAVA_HOME=/usr/java/
jdk1.6.0_45" >> /etc/profile
Building Kafka
The following steps need to be followed for building and packaging Kafka:
1. Change the current directory to the downloaded Kafka directory by using the following command:
[root@localhost opt]# cd kafka-<VERSION>
2. The directory structure for Kafka 0.8 looks as follows:
3. The following command downloads all the dependencies such as Scala compiler, Scala libraries, Zookeeper, Core-Kafka update, and Hadoop consumer/producer update, for building Kafka:
[root@localhost opt]#./sbt update
Chapter 2
[ 15 ]
On execution of the previous command, you should see the following output on the command prompt:
4. Finally, compile the complete source code for Core-Kafka, Java examples, and Hadoop producer/consumer, and package them into JAR files using the following command:
[root@localhost opt]#./sbt package
On execution of the previous command, you should see the following output on the command prompt:
5. The following additional command is only needed with Kafka 0.8 for producing the dependency artifacts:
[root@localhost opt]#./sbt assembly-package-dependency
On execution of the previous command, you should see the following output on the command prompt:
If you are planning to play with Kafka 0.8, you may experience lot of warnings with update and package commands, which can be ignored.
Summary
In this chapter we have learned how to install and build Kafka 0.7.x and 0.8. The
following chapter discusses the steps required to set up single/multibroker Kafka
clusters. From here onwards, the book only focuses on Kafka 0.8.
Setting up the Kafka Cluster
Now we are ready to play with the Apache Kafka publisher-based messaging system.
With Kafka, we can create multiple types of clusters, such as the following:
• Single node – single broker cluster
• Single node – multiple broker cluster
• Multiple node – multiple broker cluster
All the commands and cluster setups in this chapter are based on Kafka 0.8.
With Kafka 0.8, replication of clusters can also be established, which will be discussed in brief in the last part of this chapter.
So let's start with the basics.
Single node – single broker cluster
This is the starting point of learning Kafka. In the previous chapter, we built and installed Kafka on a single machine. Now it is time to setup single node – single broker based Kafka cluster as shown in the following diagram
ZooKeeper
Kafka Broker
Consumers Consumers Consumers Producers
Producers Producers
Single Node Single Kafka Broker
-
Starting the ZooKeeper server
Kafka provides the default and simple ZooKeeper configuration file used for launching a single local ZooKeeper instance. Here, ZooKeeper serves as the coordination interface between the Kafka broker and consumers. The Hadoop overview given on the Hadoop Wiki site is as follows ( http://wiki.apache.org/
hadoop/ZooKeeper/ProjectDescription ):
"ZooKeeper ( http://zookeeper.apache.org ) allows distributed processes coordinating with each other through a shared hierarchical name space of data registers (znodes), much like a file system.
The main differences between ZooKeeper and standard filesystems are that every znode can have data associated with it and znodes are limited to the amount of data that they can have. ZooKeeper was designed to store coordination data: status information, configuration, location information, and so on."
First start the ZooKeeper using the following command:
[root@localhost kafka-0.8]# bin/zookeeper-server-start.sh config/
zookeeper.properties
You should get an output as shown in the following screenshot:
Kafka comes with the required property files defining minimal properties required for a single broker – single node cluster.
The important properties defined in zookeeper.properties are shown in the following code:
# Data directory where the zookeeper snapshot is stored.
dataDir=/tmp/zookeeper
# The port listening for client request clientPort=2181
By default, the ZooKeeper server will listen on *:2181/tcp . For detailed information on
how to set up multiple servers of ZooKeeper, visit http://zookeeper.apache.org/ .
Chapter 3
[ 19 ]
Starting the Kafka broker
Now start the Kafka broker using the following command:
[root@localhost kafka-0.8]# bin/kafka-server-start.sh config/server.
properties
You should now see the output as shown in the following screenshot:
server.properties defines the following important properties required for the Kafka broker:
# The id of the broker. This must be set to a unique integer for each broker.
Broker.id=0
# The directory under which to store log files log.dir=/tmp/kafka8-logs
# Zookeeper connection string zookeeper.connect=localhost:2181
The last section in this chapter defines few more important properties available for the Kafka broker. For a complete list of Kafka broker properties, visit http://kafka.
apache.org/documentation.html#brokerconfigs .
Creating a Kafka topic
Kafka provides a command-line utility for creating topics on the Kafka server. Let's create a topic named kafkatopic with a single partition and only one replica using this utility:
[root@localhost kafka-0.8]# bin/kafka-create-topic.sh --zookeeper localhost:2181 --replica 1 --partition 1 --topic kafkatopic
You should get an output as shown in the following screenshot:
The previously mentioned utility will create a topic and show the successful creation message as shown in the previous screenshot.
Starting a producer for sending messages
Kafka provides users with a command-line producer client that accepts inputs from the command line and publishes them as a message to the Kafka cluster. By default, each new line entered is considered as a new message. The following command is used to start the console-based producer for sending the messages
[root@localhost kafka-0.8]# bin/kafka-console-producer.sh --broker-list
localhost:9092 --topic kafkatopic
Chapter 3
[ 21 ]
You should see an output as shown in the following screenshot:
While starting the producer's command-line client, the following parameters are required:
• broker-list
• topic
broker-list specifies the brokers to be connected as <node_address:port> , that is, localhost:9092 . The topic Kafkatopic is a topic that was created in the Creating a Kafka topic section. The topic name is required for sending a message to a specific group of consumers.
Now type the following message, This is single broker , and press Enter.
You should see an output as shown in the following screenshot:
Try some more messages.
Detailed information about how to write producers for Kafka and producer
properties will be discussed in Chapter 5, Writing Producers.
Starting a consumer for consuming messages
Kafka also provides a command-line consumer client for message consumption.
The following command is used for starting the console-based consumer that shows output at command line as soon as it subscribes to the topic created in Kafka broker:
[root@localhost kafka-0.8]# bin/kafka-console-consumer.sh --zookeeper localhost:2181 --topic kafkatopic --from-beginning On execution of the previous command, you should get an output as shown in the following screenshot:
The default properties for the consumer are defined in consumer.properties . The important properties are:
# consumer group id (A string that uniquely identifies a set of consumers # within the same consumer group)
groupid=test-consumer-group
# zookeeper connection string zookeeper.connect=localhost:2181
Detailed information about how to write consumers for Kafka and consumer properties is discussed in Chapter 6, Writing Consumers.
By running all four components ( zookeeper , broker , producer , and consumer ) in different terminals, you will be able to enter messages from the producer's terminal and see them appearing in the subscribed consumer's terminal.
Usage information for both producer and consumer command-line tools can
be viewed by running the command with no arguments.
Chapter 3
[ 23 ]
Single node – multiple broker cluster
Now we have come to the next level of Kafka cluster. Let us now set up single node – multiple broker based Kafka cluster as shown in the following diagram:
Producers
ZooKeeper
Consumers
Consumers
Consumers Producers
Producers
Single Node Single Kafka Broker - Broker 1
Broker 2
Broker 3
Starting ZooKeeper
The first step of starting ZooKeeper remains the same for this type of cluster.
Starting the Kafka broker
For setting up multiple brokers on a single node, different server property files are required for each broker. Each property file will define unique, different values for the following properties:
• brokerid
• port
• log.dir
For example, in server-1.properties used for broker1 , we define the following:
• brokerid=1
• port=9092
• log.dir=/tmp/kafka8-logs/broker1
Similarly, for server-2.properties used for broker2 , we define the following:
• brokerid=2
• port=9093
• log.dir=/tmp/kafka8-logs/broker2
A similar procedure is followed for all brokers. Now, we start each broker in a separate console using the following commands:
[root@localhost kafka-0.8]# env JMX_PORT=9999 bin/kafka-server-start.sh config/server-1.properties
[root@localhost kafka-0.8]# env JMX_PORT=10000 bin/kafka-server-start.sh config/server-2.properties
Similar commands are used for all brokers You will also notice that we have defined a separate JMX port for each broker.
The JMX ports are used for optional monitoring and troubleshooting with tools such as JConsole.
Creating a Kafka topic
Using the command-line utility for creating topics on the Kafka server, let's create a topic named othertopic with two partitions and two replicas:
[root@localhost kafka-0.8]# bin/kafka-create-topic.sh --zookeeper localhost:2181 --replica 2 --partition 2 --topic othertopic
Starting a producer for sending messages
If we use a single producer to get connected to all the brokers, we need to pass the initial list of brokers, and the information of the remaining brokers is identified by querying the broker passed within broker-list , as shown in the following command. This metadata information is based on the topic name.
--broker-list localhost:9092,localhost:9093 Use the following command to start the producer:
[root@localhost kafka-0.8]# bin/kafka-console-producer.sh --broker-list localhost:9092,localhost:9093 --topic othertopic
If we have a requirement to run multiple producers connecting to different
combinations of brokers, we need to specify the broker list for each producer
like we did in the case of multiple brokers.
Chapter 3
[ 25 ]
Starting a consumer for consuming messages
The same consumer client, as in the previous example, will be used in this process.
Just as before, it shows the output on the command line as soon as it subscribes to the topic created in the Kafka broker:
[root@localhost kafka-0.8]# bin/kafka-console-consumer.sh --zookeeper localhost:2181 --topic othertopic --from-beginning
Multiple node – multiple broker cluster
This cluster scenario is not discussed in detail in this book, but as in the case of multiple-node Kafka cluster, where we set up multiple brokers on each node, we should install Kafka on each node of the cluster, and all the brokers from the different nodes need to connect to the same ZooKeeper.
For testing purposes, all the commands will remain identical to the ones we used in the single node – multiple brokers cluster.
The following diagram shows the cluster scenario where multiple brokers are configured on multiple nodes (Node 1 and Node 2 in this case), and the producers and consumers are getting connected in different combinations:
Producers
ZooKeeper
Consumers
Consumers
Consumers Producers
Producers
Single Node Single Kafka Broker -
Broker 1
Broker 2
Node 1
Broker 1
Broker 2
Node 2
Kafka broker property list
The following is the list of few important properties that can be configured for the Kafka broker. For the complete list, visit http://kafka.apache.org/
documentation.html#brokerconfig .
Property name Description Default value
broker.id Each broker is uniquely identified by an ID.
This ID serves as the broker's name, and allows the broker to be moved to a different host/port without confusing consumers.
0
log.dirs These are the directories in which the log
data is kept. /tmp/kafka-
logs zookeeper.connect This specifies the ZooKeeper's connection
string in the form hostname:port/
chroot. Here, chroot is a base directory that is prepended to all path operations (this effectively namespaces all Kafka znodes to allow sharing with other applications on the same ZooKeeper cluster).
localhost:2181
Summary
In this chapter, we have learned how to set up a Kafka cluster with single/multiple brokers on a single node, run command-line producers and consumers, and
exchange some messages. We have also discussed some details about setting up a multinode – multibroker cluster.
In the next chapter, we will look at the internal design of Kafka.
Kafka Design
Before we start getting our hands dirty by coding Kafka producers and consumers, let's quickly discuss the internal design of Kafka.
In this chapter we shall be focusing on the following topics:
• Kafka design fundamentals
• Message compression in Kafka
• Cluster mirroring in Kafka
• Replication in Kafka
Due to the overheads associated with JMS and its various implementations and limitations with the scaling architecture, LinkedIn ( www.linkedin.com ) decided to build Kafka to address their need for monitoring activity stream data and operational metrics such as CPU, I/O usage, and request timings.
While developing Kafka, the main focus was to provide the following:
• An API for producers and consumers to support custom implementation
• Low overhead for network and storage with message persistence
• High throughput supporting millions of messages
• Distributed and highly scalable architecture
Kafka design fundamentals
In a very basic structure, a producer publishes messages to a Kafka topic, which is created on a Kafka broker acting as a Kafka server. Consumers then subscribe to the Kafka topic to get the messages. This is described in the following diagram:
Message Producers
Producers
Single Node Single Kafka Broker -
Topic
Consumer Group Consumers Consumers Mes sage
Message
In the preceding diagram a single node – single broker architecture is shown.
This architecture considers that all three parties—producers, Kafka broker, and consumers—are running on different machines.
Here, each consumer is represented as a process and these processes are organized within groups called consumer groups.
A message is consumed by a single process (consumer) within the consumer group, and if the requirement is such that a single message is to be consumed by multiple consumers, all these consumers need to be kept in different consumer groups.
By Kafka design, the message state of any consumed message is maintained within the message consumer, and the Kafka broker does not maintain a record of what is consumed by whom, which also means that poor designing of a custom consumer ends up in reading the same message multiple times.
Important Kafka design facts are as follows:
• The fundamental backbone of Kafka is message caching and storing it on the filesystem. In Kafka, data is immediately written to the OS kernel page.
Caching and flushing of data to the disk is configurable.
• Kafka provides longer retention of messages ever after consumption, allowing consumers to reconsume, if required.
• Kafka uses a message set to group messages to allow lesser network overhead.
Chapter 4
[ 29 ]
• Unlike most of the messaging systems, where metadata of the consumed messages are kept at server level, in Kafka, the state of the consumed
messages is maintained at consumer level. This also addresses issues such as:
° Loosing messages due to failure
° Multiple deliveries of the same message
By default, consumers store the state in ZooKeeper, but Kafka also allows storing it within other storage systems used for Online Transaction Processing (OLTP) applications as well.
• In Kafka, producers and consumers work on the traditional push-and-pull model, where producers push the message to a Kafka broker and consumers pull the message from the broker.
• Kafka does not have any concept of a master and treats all the brokers as peers. This approach facilitates addition and removal of a Kafka broker at any point, as the metadata of brokers are maintained in ZooKeeper and shared with producers and consumers.
• In Kafka 0.7.x, ZooKeeper-based load balancing allows producers to discover the broker dynamically. A producer maintains a pool of broker connections, and constantly updates it using ZooKeeper watcher callbacks. But in
Kafka 0.8.x, load balancing is achieved through Kafka metadata API and ZooKeeper can only be used to identify the list of available brokers.
• Producers also have an option to choose between asynchronous or synchronous mode for sending messages to a broker.
Message compression in Kafka
As we have discussed, Kafka uses message set feature for grouping the messages.
It also provides a message group compression feature. Here, data is compressed by the message producer using either GZIP or Snappy compression protocols and decompressed by the message consumer. There is lesser network overhead for the compressed message set where it also puts very little overhead of decompression at the consumer end.
This compressed set of messages can be presented as a single message to the consumer who later decompresses it. Hence, the compressed message may have infinite depth of messages within itself.
To differentiate between compressed and uncompressed messages, a compression-
attributes byte is introduced in the message header. Within this compression byte,
the lowest two bits are used to represent the compression codec used for compression
and the value 0 of these last two bits represents an uncompressed message.
Message compression techniques are very useful for mirroring the data across datacenters using Kafka, where large amounts of data get transferred from active to passive datacenters in compressed format.
For more detailed usage on Kafka message compression, visit https://
cwiki.apache.org/confluence/display/KAFKA/Compression.
Cluster mirroring in Kafka
The Kafka mirroring feature is used for creating the replica of an existing cluster, for example, for the replication of an active datacenter into a passive datacenter.
Kafka provides a mirror maker tool for mirroring the source cluster into target cluster.
The following diagram depicts the mirroring tool placement in architectural form:
Source Single Node Cluster Producers
Data
ZooKeeper
Kafka Broker
ZooKeeper
Kafka Broker
Consumer Producer
Re-Publish
Mirror Maker Target Single Node
Cluster
Consumer
Data
In this architecture, the job of the mirror tool is to consume the messages from the source cluster and republish them on the target cluster.
A similar approach is used by the Kafka migration tool to migrate from 0.7.x Kafka cluster to 0.8.x Kafka cluster.
For detailed explanation on the mirror maker tool setup and configuration, visit https://cwiki.apache.org/confluence/
display/KAFKA/Kafka+mirroring+%28MirrorMaker%29.
Chapter 4
[ 31 ]
Replication in Kafka
Before we talk about replication in Kafka, let's talk about message partitioning.
In Kafka, message partitioning strategy is used at the Kafka broker end. The decision about how the message is partitioned is taken by the producer, and the broker stores the messages in the same order as they arrive. The number of partitions can be configured for each topic within the Kafka broker.
Kafka replication is one of the very important features introduced in Kafka 0.8.
Though Kafka is highly scalable, for better durability of messages and high availability of Kafka clusters, replication guarantees that the message will be published and consumed even in case of broker failure, which may be caused by any reason. Here, both producers and consumers are replication aware in Kafka.
The following diagram explains replication in Kafka:
Multi Broker Kafka Cluster
Producers
Consumers
1 2 3 4
Kafka Topic with 4 Partition Message
Broker 1 Topic
Lead 1
Broker 2 Topic
Broker 3 Topic
Reap 1
Broker 4 Topic
Reap 1
Message
4 1
Let's discuss the preceding diagram in detail.
In replication, each partition of a message has n replicas and can afford n-1 failures
to guarantee message delivery. Out of the n replicas, one replica acts as the lead
replica for the rest of the replicas. ZooKeeper keeps the information about the lead
replica and the current in-sync follower replica (lead replica maintains the list
of all in-sync follower replicas).
Each replica stores its part of the message in local logs and offsets, and is periodically synced to the disk. This process also ensures that either a message is written to all the replicas or to none of them.
If the lead replica fails, either while writing the message partition to its local log or before sending the acknowledgement to the message producer, a message partition is resent by the producer to the new lead broker.
The process of choosing the new lead replica is that all followers' In-sync Replicas (ISRs) register themselves with ZooKeeper. The very first registered replica becomes the new lead replica, and the rest of the registered replicas become the followers.
Kafka supports the following replication modes:
• Synchronous replication: In synchronous replication, a producer first identifies the lead replica from ZooKeeper and publishes the message.
As soon as the message is published, it is written to the log of the lead replica and all the followers of the lead start pulling the message, and by using a single channel, the order of messages is ensured. Each follower replica sends an acknowledgement to the lead replica once the message is written to its respective logs. Once replications are complete and all expected acknowledgements are received, the lead replica sends an acknowledgement to the producer.
On the consumer side, all the pulling of messages is done from the lead replica.
• Asynchronous replication: The only difference in this mode is that as soon as a lead replica writes the message to its local log, it sends the acknowledgement to the message client and does not wait for the acknowledgements from follower replicas. But as a down side, this mode does not ensure the message delivery in case of broker failure.
For detailed explanation on Kafka replication and its usage, visit https://cwiki.apache.org/confluence/display/KAFKA/
Kafka+Replication.
Summary
In this chapter, we have learned the design concepts used for building a solid foundation for Kafka.
In the next chapter, we shall be focusing on how to write Kafka producers using
the API provided.
Writing Producers
Producers are applications that create messages and publish them to the Kafka broker for further consumption. These producers can be different in nature; for example, frontend applications, backend services, proxy applications, adapters to legacy systems, and producers for Hadoop. These producers can also be implemented in different languages such as Java, C, and Python.
In this chapter we shall be focusing on the following topics:
• The Kafka API for message producers
• Simple Java-based Kafka producers
• Java-based Kafka producers using message partitioning
At the end of the chapter, we will explore some of the important properties required for the Kafka producer.
Let's begin. The following diagram explains the Kafka API for message producers:
Create Message
Message Publish Message Data
Producer
