Book Image

Mastering Hadoop 3

By : Chanchal Singh, Manish Kumar
Book Image

Mastering Hadoop 3

By: Chanchal Singh, Manish Kumar

Overview of this book

Apache Hadoop is one of the most popular big data solutions for distributed storage and for processing large chunks of data. With Hadoop 3, Apache promises to provide a high-performance, more fault-tolerant, and highly efficient big data processing platform, with a focus on improved scalability and increased efficiency. With this guide, you’ll understand advanced concepts of the Hadoop ecosystem tool. You’ll learn how Hadoop works internally, study advanced concepts of different ecosystem tools, discover solutions to real-world use cases, and understand how to secure your cluster. It will then walk you through HDFS, YARN, MapReduce, and Hadoop 3 concepts. You’ll be able to address common challenges like using Kafka efficiently, designing low latency, reliable message delivery Kafka systems, and handling high data volumes. As you advance, you’ll discover how to address major challenges when building an enterprise-grade messaging system, and how to use different stream processing systems along with Kafka to fulfil your enterprise goals. By the end of this book, you’ll have a complete understanding of how components in the Hadoop ecosystem are effectively integrated to implement a fast and reliable data pipeline, and you’ll be equipped to tackle a range of real-world problems in data pipelines.
Table of Contents (23 chapters)
Title Page
Dedication
About Packt
Foreword
Contributors
Preface
Index

Serialization


Serialization is the process of converting structured objects into a byte stream that will be transferred over a network or will be written to a persistent storage. Deserialization is the process of converting a byte stream back into structured objects. The basic question that some of us always have is, why do we need serialization? Let us understand it in simple terms. Every language or application has its own way of representing data, for example, Java has objects to represent data, Spark has RDD to represent data, MapReduce has writable objects to represent data, and so on. These representations are only known to frameworks that can be processed in memory, but this data cannot be shared between different processes or applications that have a different way of representing data. Now, we are clear that data needs some common representation when it is written to the storage system or shared across networks to be used by different applications. In most cases, writing data into...