Book Image

Serverless Design Patterns and Best Practices

By : Brian Zambrano
Book Image

Serverless Design Patterns and Best Practices

By: Brian Zambrano

Overview of this book

Serverless applications handle many problems that developers face when running systems and servers. The serverless pay-per-invocation model can also result in drastic cost savings, contributing to its popularity. While it's simple to create a basic serverless application, it's critical to structure your software correctly to ensure it continues to succeed as it grows. Serverless Design Patterns and Best Practices presents patterns that can be adapted to run in a serverless environment. You will learn how to develop applications that are scalable, fault tolerant, and well-tested. The book begins with an introduction to the different design pattern categories available for serverless applications. You will learn thetrade-offs between GraphQL and REST and how they fare regarding overall application design in a serverless ecosystem. The book will also show you how to migrate an existing API to a serverless backend using AWS API Gateway. You will learn how to build event-driven applications using queuing and streaming systems, such as AWS Simple Queuing Service (SQS) and AWS Kinesis. Patterns for data-intensive serverless application are also explained, including the lambda architecture and MapReduce. This book will equip you with the knowledge and skills you need to develop scalable and resilient serverless applications confidently.
Table of Contents (18 chapters)
Title Page
Copyright and Credits
Dedication
Packt Upsell
Contributors
Preface
Index

Asynchronous processing of Twitter streams


Twitter is an excellent source of random data. Given the volume and variety of data, we can readily come up with example (and real) problems to solve. In our case, we're going to build a serverless processing system by sipping off the public twitter stream. Our example system will  have the following workflow:

  1. Read a tweet with cat or dog images from the Twitter firehose
  2. Place messages on an SQS queue.
  3. Worker processes will read those image URLs off the queue and perform image recognition.

While this example can be a bit contrived, the concepts demonstrated are true to life. We'll use the AWS Rekognition service to perform image recognition and labeling of any cat or dog images we find. Rekognition is quite fast at what it does, but it's easy to imagine processing images with a much slower service. In that case, adding items onto a queue and processing them at our leisure with one or more worker processes would allow us to scale out to achieve a higher...