Book Image

Simplify Big Data Analytics with Amazon EMR

By : Sakti Mishra
Book Image

Simplify Big Data Analytics with Amazon EMR

By: Sakti Mishra

Overview of this book

Amazon EMR, formerly Amazon Elastic MapReduce, provides a managed Hadoop cluster in Amazon Web Services (AWS) that you can use to implement batch or streaming data pipelines. By gaining expertise in Amazon EMR, you can design and implement data analytics pipelines with persistent or transient EMR clusters in AWS. This book is a practical guide to Amazon EMR for building data pipelines. You'll start by understanding the Amazon EMR architecture, cluster nodes, features, and deployment options, along with their pricing. Next, the book covers the various big data applications that EMR supports. You'll then focus on the advanced configuration of EMR applications, hardware, networking, security, troubleshooting, logging, and the different SDKs and APIs it provides. Later chapters will show you how to implement common Amazon EMR use cases, including batch ETL with Spark, real-time streaming with Spark Streaming, and handling UPSERT in S3 Data Lake with Apache Hudi. Finally, you'll orchestrate your EMR jobs and strategize on-premises Hadoop cluster migration to EMR. In addition to this, you'll explore best practices and cost optimization techniques while implementing your data analytics pipeline in EMR. By the end of this book, you'll be able to build and deploy Hadoop- or Spark-based apps on Amazon EMR and also migrate your existing on-premises Hadoop workloads to AWS.
Table of Contents (19 chapters)
1
Section 1: Overview, Architecture, Big Data Applications, and Common Use Cases of Amazon EMR
6
Section 2: Configuration, Scaling, Data Security, and Governance
11
Section 3: Implementing Common Use Cases and Best Practices

Test your knowledge

Before moving on to the next chapter, test your knowledge with the following questions:

  1. Assume your data science team is using EMR notebooks for their interactive development and they are primarily using Python 3 for machine learning model development. When they started executing the Python code, they found some of their scripts are not getting executed; they get an error stating that the Python module does not exist. How would you make the additional Python modules available in the EMR notebook so that your data science team can continue executing their scripts for machine learning model development?
  2. Assume you have an S3 data lake on top of which you have created Hudi tables for ACID transactions and UPSERT. You are updating records as they change, which creates multiple versions of the records in the Hudi table. You have received a business requirement to find the value of a specific column at a specific time. How would you fulfill that requirement...