Book Image

Data Engineering with AWS

By : Gareth Eagar
Book Image

Data Engineering with AWS

By: Gareth Eagar

Overview of this book

Written by a Senior Data Architect with over twenty-five years of experience in the business, Data Engineering for AWS is a book whose sole aim is to make you proficient in using the AWS ecosystem. Using a thorough and hands-on approach to data, this book will give aspiring and new data engineers a solid theoretical and practical foundation to succeed with AWS. As you progress, you’ll be taken through the services and the skills you need to architect and implement data pipelines on AWS. You'll begin by reviewing important data engineering concepts and some of the core AWS services that form a part of the data engineer's toolkit. You'll then architect a data pipeline, review raw data sources, transform the data, and learn how the transformed data is used by various data consumers. You’ll also learn about populating data marts and data warehouses along with how a data lakehouse fits into the picture. Later, you'll be introduced to AWS tools for analyzing data, including those for ad-hoc SQL queries and creating visualizations. In the final chapters, you'll understand how the power of machine learning and artificial intelligence can be used to draw new insights from data. By the end of this AWS book, you'll be able to carry out data engineering tasks and implement a data pipeline on AWS independently.
Table of Contents (19 chapters)
1
Section 1: AWS Data Engineering Concepts and Trends
6
Section 2: Architecting and Implementing Data Lakes and Data Lake Houses
13
Section 3: The Bigger Picture: Data Analytics, Data Visualization, and Machine Learning

Extending analytics with data warehouses/data marts

Tools such as Amazon Athena (which we will do a deeper dive into in Chapter 11, Ad Hoc Queries with Amazon Athena) allow us to run SQL queries directly on data in the data lake. And while this enables us to query very large datasets that exist on Amazon S3, the performance of these queries is generally lower than the performance you get when running queries against data on a high-performance disk that is local to the compute engine.

Not all queries require this kind of high performance, and we can categorize our queries and data into three categories. Let's take a look.

Cold data

This is data that is not frequently accessed, but it is mandatory to store it for long periods for compliance and governance reasons, or historical data that is stored to enable future research and development (such as for training machine learning models).

An example of this is the logs from a banking website. Unless there is a breach...