Book Image

Scalable Data Architecture with Java

By : Sinchan Banerjee
Book Image

Scalable Data Architecture with Java

By: Sinchan Banerjee

Overview of this book

Java architectural patterns and tools help architects to build reliable, scalable, and secure data engineering solutions that collect, manipulate, and publish data. This book will help you make the most of the architecting data solutions available with clear and actionable advice from an expert. You’ll start with an overview of data architecture, exploring responsibilities of a Java data architect, and learning about various data formats, data storage, databases, and data application platforms as well as how to choose them. Next, you’ll understand how to architect a batch and real-time data processing pipeline. You’ll also get to grips with the various Java data processing patterns, before progressing to data security and governance. The later chapters will show you how to publish Data as a Service and how you can architect it. Finally, you’ll focus on how to evaluate and recommend an architecture by developing performance benchmarks, estimations, and various decision metrics. By the end of this book, you’ll be able to successfully orchestrate data architecture solutions using Java and related technologies as well as to evaluate and present the most suitable solution to your clients.
Table of Contents (19 chapters)
1
Section 1 – Foundation of Data Systems
5
Section 2 – Building Data Processing Pipelines
11
Section 3 – Enabling Data as a Service
14
Section 4 – Choosing Suitable Data Architecture

Hybrid data processing patterns

In this section, we will discuss two very famous patterns that support both batch and real-time processing. Since these patterns support both batch processing and stream processing, they are categorized as hybrid patterns. Let’s take a look at the most popular hybrid architectural patterns.

The Lambda architecture  

First, let’s understand the need for Lambda architecture. In distributed computing, the CAP theorem states that any distributed data can guarantee only two out of the three features of the data – that is, consistency, availability, and partition tolerance. However, Nathan Marz proposed a new pattern in 2011 that made it possible to have all three characteristics present in a distributed data store. This pattern is called the Lambda pattern. The Lambda architecture consists of three layers, as follows:

  • Batch layer: This layer is responsible for batch processing
  • Speed layer: This layer is responsible...