Truly greenfield development of microservices-based applications is relatively rare. Many organizations that want to adopt microservices already have a monolithic application. As I describe in chapter 13 of my book, the recommended approach is to use the Strangler application pattern and incrementally migrate function from the monolith into services.
These presentations are good overviews:
The FTGO monolith (see source code) is the monolithic version of the microservices-based FTGO application. Its goal is to demonstrate how to refactor a monolithic application into services using the Stranger pattern. It’s used as an example in both my book and in my presentations about refactoring.
Chapter 13 of my book describes how to refactor a monolith to microservices. There are two main refactoring strategies:
Let’s look at each one.
A good way to begin the migration to microservices is to implement significant new functionality as services. This is sometimes easier than breaking apart of the monolith. It also demonstrates to the business that using microservices significantly accelerates software delivery.
While implementing new functionality as services is extremely useful, the only way of eliminating the monolith is to incrementally extract modules out of the monolith and convert them into services.
For example, let’s imagine that the FTGO team wants to improve the efficiency of the business and customer satisfaction by rapidly iterating on the courier scheduling algorithm.
It will be a lot easier for them to work on the delivery management logic if it’s a separate
To do that, the FTGO team must separate delivery management from order management and convert it into service.
Delivery Service consists of the following the steps:
Before looking at each of these steps, let’s first review the AS-IS code.
In the FTGO monolithic application,
Order management and
Delivery management are intertwined as shown in this diagram.
Read more about the AS-IS architecture.
The first step is to split the code and convert delivery management into a separate, loosely coupled module within the monolith.
Read more about how the code is split.
The second step is split the database and define a separate database schema for the
Read more about splitting the database.
The third step of the refactoring process is to define a standalone
Delivery Service and deploy it.
The service does not, however, handle production traffic.
Instead, it can, for example, be tested in production.
Read more about defining standalone
The fourth step of the refactoring process is to use the standalone
Read more about how to route production traffic to the
The fifth step of the refactoring process is to remove the now obsolete delivery management logic from the monolith.
Read more about removing the obsolete code.
Signup to get notified when more content is added to this page.