Microservice Architecture Best Practices - Messaging Queues


In this article, we discuss why queues are needed, and how they form the cornerstone of asynchronous communication in microservices architectures.

What you will learn

  • What is a Queue?
  • What is asynchronous communication or asynchronous messaging?
  • What are the advantages of using Queues in microservices architectures?

Best Practices with Cloud and Microservices

This is the fourth article in a series of six articles on best practices with cloud and microservices:

Why Asynchronous Messaging?

Why is asynchronous messaging important?

Consider the simple example of an ordering service:

image info

A customer places an order through the OrderService. Let’s say following steps are involved:

  • It needs to send a request to the StockService
  • Send communication to the user through the EmailService and SMSService
  • Call the PackageService to start delivery

Option 1 : Single Component for all processing

One way to design this application would be to have a single component that accomplishes all this functionality. For example, write a Java class which accepts the order, and does the entire processing itself.

Option 2 : Introducing Asynchronous Messaging using Queues

The other option is to have a queue in-between the receiving OrderService component, and the rest of the components. When the OrderService receives an order, it places the request on the Queue. The rest of the components are independent services, listening on the Queue.

As soon as an order is placed on the queue:

  • The StockService processes the order and updates the database
  • The EmailService sends out an email to the customer
  • The SMSService sends out an SMS to the user
  • The PackageService does the required package processing on the item

Single Component v Queue

If your business goal is to support a few hundreds or thousands of users, then a simple architecture would be a good choice. The first architecture might be all you need.

However, when we talk of a large scale e-commerce web application such as Amazon.com, that receives millions of orders in a short time, you need a lot of flexibility. That’s when you go for the second approach, of bringing a queue in.

Advantages of Using A Messaging Queue

Let’s look at the advantages of using asynchronous communication based on a messaging queue.

A Queue Improves Reliability

Let’s say the SMSService is down for a short time.

In Option 1, since the OrderService directly invokes the SMSService on receiving an order, an SMS cannot be sent out. That might mean canceling the order, as all steps are part of a single transaction. Failure of one of the components would lead to cancellation of the customer request, with a need for him to re-initiate the order - at a later point in time.

However in the second approach, called asynchronous communication, the order request is placed on the queue. When the SMSService comes back up, it will find the order event and process it. It can then send out the SMS messages for all its pending requests.

A Queue Provides Scalability

Suppose on a given day, there is a need to send out 100,000 SMS messages. An option with having asynchronous communication is that you can increase the number of instances of the SMSService. This flexibility in number of instances of the components, improves system scalability.

A Queue Improves Testability

Each of the components/services are loosely tied, and have independent requirements. It is thus much easier to test each one of them, preferably in isolation.

A Queue Improves Maintainability

A queue also improves system maintainability, as the application is divided into smaller services.

A Queue Improves Flexibility

Not only increase instances of existing services, a queue makes the system flexible to add more services in future. All that you need to do is add a new service listen to events on the queue and consuming them. The new service could be plugged in, with a minimum of down time.

An extremely popular message queue framework is RabbitMQ.

Different frameworks use different communication protocols, but the concept underlying all of them is the same. At the core, a queue separates the component that generates the event, from the services that consume the event.

In general, if you need a system that addresses a very large user base, whose users submit a large number of requests to be processed, go for queue-based systems.

Do check out our video on this:

image info

Summary

In this article, we looked at the fact that there are two types of architectures - synchronous and asynchronous. Synchronous systems have the disadvantage that one component being down, causes loss of service to the entire system. Asynchronous systems solve this problem by introducing an intermediary queue to hold events. An architecture based on asynchronous communication using a queue improves testability, scalability, maintainability and flexibility.

Congratulations! You are reading an article from a series of 50+ articles on Spring, Spring Boot , Hibernate, Full Stack, Cloud and Microservices. We also have 20+ projects on our Github repository. For the complete series of 50+ articles and code examples, click here.

Join 300,000 Learners!

Learn Spring Boot in 10 Steps - FREE Course

Next Steps

Image

Image

Image Image

Image

Image

Image

Image

Image

Image

Image

Image

Image

Image

Image

Image

Related Posts

Spring Boot Tutorials for Beginners

At in28Minutes, we are creating a number of tutorials with videos, articles & courses on Spring Boot for Beginners and Experienced Developers. This resources will help you learn and gain expertise at Spring Boot.

Spring and Spring Boot Video Tutorials for Beginners

At in28Minutes, we are creating a number of tutorials with videos, articles & courses on Spring Boot for Beginners and Experienced Developers. Here's a list of video tutorials and courses for you

Software Design - Separation Of Concerns - with examples

Software architects and programmers love having Seperation of Concerns. What is it? Why is it important? Let's get started.

Object Oriented Software Design - Solid Principles - with examples

Software design is typically complex. Object oriented design takes it to the next level. There are a number of design patterns and other stuff to be aware of. Can we make things simple? What are the goals to aim for when you are doing object oriented design? SOLID Principles is a great starting point for Object Oriented Design.

Software Design - Open Closed Principle - with examples

Open Closed Principle is one of the SOLID Principles. You want your code to be easily extended. How do you achieve it with minimum fuss? Let's get started.

Software Design - What is Dependency Inversion Principle?

Dependency Inversion Principle is one of the important SOLID Principles. Dependency Inversion Principle is implemented by one of the most popular Java frameworks - Spring. What is it all about? How does it help you design good applications?

Introduction to Four Principles Of Simple Design

With agile and extreme programming, the focus is on keeping your design simple. How do you keep your design simple? How do you decide whether your code is good enough?

Software Design - Single Responsibility Principle - with examples

For me, Single Responsibility Principle is the most important design principle. What is Single Responsibility Principle? How do you use it? How does it help with making your software better? Let's get started.

REST API Best Practices - With Design Examples from Java and Spring Web Services

Designing Great REST API is important to have great microservices. How do you design your REST API? What are the best practices?

Designing REST API - What is Code First Approach?

Designing Great REST API is important to have great microservices. Code First approach focuses on generating the contract from code. Is it the best possible approach?