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

Data model design considerations

In this section, we will briefly discuss various design considerations you should consider while designing a data model for the various databases discussed in the previous section. The following aspects need to be considered while designing a data model:

  • Normalized versus denormalized: Normalization is a data organization technique. It is used to reduce redundancy in a relationship or set of relationships. This is highly used in RDBMS, and it is always a best practice in RDBMS to create a normalized data model. In a normalized data model, you store a column in one of the tables (which is most suitable), rather than storing the same column in multiple tables. When fetching data, if you need the data of that column, you can join the tables to fetch that column. The following diagram shows an example of normalized data modeling using the crows-feet notation:

Figure 2.11 – Normalized data modeling

In the preceding...