Introduction to MVC concepts and their implementation in Rails.
Clone or download

General Assembly Logo

Rails API Overview


By the end of this, developers should be able to:

  • Explain why an API is necessary.
  • List some of the responsibilities of a typical API, and identify which components within MVC map to those responsibilities.
  • Diagram the process flow from Client through Database and back
  • Map MVC roles to specific components of Rails.
  • Indicate where different types of files can be found within a Rails application.


  1. Fork and clone this repository. FAQ
  2. Create a new branch, training, for your work.
  3. Checkout to the training branch.
  4. Install dependencies with bundle install.


Who Needs An API

By now, you've built a client application: it had a user interface (UI) and behavioral logic, allowing users to play Tic-Tac-Toe in the browser. Isn't that enough? Why might we want an API?

One possible reason is that when two different people are served up a web page, they each have a separate copy of everything on the page. As a result, if we have two users interacting with our app, they can never interact with each other, or even be aware of each other.

Another possible reason is that we might want want to store the state of the game over time. What if we wanted to be able to 'save' our game, close the browser, walk away, and pick it up again the next day? Having an API where data is stored means that we can do just that. This is known as persistence (things being saved over time).

For your client, we provided an API that could do all that. But how did it work?

Lab - Design Your Own API

In your teams, take ten minutes (and a patch of whiteboard), and write out all the different things that you think that the API for your Tic-Tac-Toe clients needed to do in order to work correctly.

Then, once you've done this, take a second look at what you've written down. Assuming you were to try to build a Ruby program to handle all of these different tasks, what abstractions would you need? How would the different responsibilities be broken up? Make a list. It doesn't matter how "correct" it is. For this part, recall our strategies in modeling. What are the relevant nouns and verbs? Big ideas and details?

Once all groups are finished, we'll discuss our results.

One Possible Solution - Model-View-Controller

Suppose we wanted to build an API app that records a user's height and weight. It might work as follows:

  1. A user interacts with a front-end application, triggering a POST, and this POST request contains data - specifically, the latest measurements of height and weight.
  2. When the POST request is received by the server, the API app parses the request and extracts the relevant information.
  3. The data from the POST request gets added to our records.
  4. As confirmation, a JSON with a unique identifier (so we can refer to this specific measurement in the future) gets sent to the front end.

Our app can also retrieve records, like so:

  1. Something on the front-end (probably driven by user interaction) triggers a GET request that asks for a specific record.
  2. When the GET request is received by the server, the API app parses the request and identifies which measurement record is being requested.
  3. The data for the desired measurement gets retrieved from our set of records.
  4. Finally, the desired data gets put into a JSON and sent back to the front end.

If we were to try to generalize and abstract away the differences between these two steps, we might say that our web application needs to:

  • Receive incoming requests from a front-end.
  • Execute specific behaviors in response to those requests.
  • Create, read, update, or destroy data records through some kind of data storage system.
  • Share information back to the browser.

The quartet of 'create', 'read', 'update', and 'destroy' is commonly known as 'CRUD'; each refers to a specific type of action that can be performed on our data storage system. Each of these types might have more than one specific action associated with it ('read one' vs 'read all', for instance).

One common way of dividing up these four responsibilities is the Model-View-Controller (MVC) architecture pattern. This pattern involves making three core types of components, each responsible for a different part of the API's functionality.

A Model directly manages the data in our application, and provides a representation of that data for the rest of the application to use.

A View is like it sounds - it's data that gets sent back to the client for the user to consume.

A Controller responds to user requests as they come in, and utilizes both the model and the view components to perform the desired behavior and produce a response.

In addition to these three types of components, however, there is a fourth piece that it's important to consider with web development particularly: routing. Routes indicate to the server which controllers should be triggered (and how) by which kinds of requests. It's a critical piece of the puzzle, and one we'll be looking at later today in more detail.

What which part(s) of an HTTP request does the router use to determine which code to run?

MVC architecture is very common in web applications, and Rails gives us the tools to spin up applications that are roughly in line with the idea of MVC.

Lab - Act Out an MVC API

We're going to act out the various parts of an MVC application. Link up with another team so that you're in a group of eight, and assign each member of your 'super-group' one of the roles below. These roles are:

  • Client
  • Server
    • Router
    • Controllers
      • People Controller
      • Places Controller
      • Things Controller
    • Models
      • People Model
      • Places Model
      • Things Model

Once you've divvied up the roles, take two minutes to read through the directions for your role, found roles directory. These directions explain what your responsibilities are and how you should carry them out. However, you may only communicate with the teammates listed in your role card.

  • Developers playing the role of the model should record their information in the tab associated with their group number found at the bottom of the spreadsheet link in ./roles/model.

We'll work through one example request together, and then each group will work independently to answer all remaining requests. Once all groups have finished, we'll have a retro-style discussion about how everything went.

MVC with Rails

Rails is a web framework - a tool that helps us quickly and easily build web applications - written in Ruby, and designed and created by David Heinemeier Hannson (also known as 'DHH'). Although Rails applications don't match up perfectly with the abstract idea of MVC, their architecture is fairly similar.

Let's take a look at an actual Rails app and see how it lines up. Fork and clone this repo. Recognize it? It's the Tic-Tac-Toe API!

Open it up in Atom. Have a look at the file structure. Let's take a step back and just look at directories at the top level of the app.

├── app
├── config
└── db

For now, you only need to think about app, config, and db - you probably won't be touching any code outside of those three directories. How is that possible? Because Rails actually builds out most of these files and folders for you, every time you use it to create a new application. That's why it's called a 'framework' - it gives you the skeleton for a brand new app, which you can then customize.

  • The app directory holds the code for our application itself. We'll be writing a lot of code here.
  • config holds configuration settings for our app and for the things that plug into it. This includes things like environmental variable settings and secret keys, but also things like the routing configuration for our server (which is not strictly part of our Rails app, but which our app uses); the routes.rb file, in particular, defines all of the routing for our app.
  • db holds files related to the structure of the application's database. The database, like the server, is separate from Rails and is not strictly part of the app, so it makes sense to keep this outside of the app directory.

Let's dive into the app directory. This app in particular has more going on than yours probably will, but it still has all the basic components.

├── controllers
├── models
└── serializers

Three of these directories should jump out at you: controllers, models, and views. Each holds the different Ruby files that Rails uses to handle the respective responsibilities of MVC.

Don't worry about assets, views, mailers, or helpers for now, as these all deal with more of the "views" portion of an application, which our front-end is handling. In API-style applications, serializers help package data to be sent to the client and replace all the views that we care about for this program. In contrast, you could build an application that handles views as well, so that your backend and frontend live together in one application. Building a web application with Rails' default settings will give you this monolithic application style.

How to Approach Understanding Rails

There is a lot of structure and code automatically provided by Rails, and as Rails developers, we only need interact with and understand a small part of it. In fact, the creator of Rails has himself stated that he doesn't need to know or understand all of the Rails framework (watch the video clip)!

For the aspects of the Rails framework we are concerned with, excellent descriptions and code examples can be found in the Getting Started with Rails guide.

Additional Resources


  1. All content is licensed under a CC­BY­NC­SA 4.0 license.
  2. All software code is licensed under GNU GPLv3. For commercial use or alternative licensing, please contact