REST was developed and formalized by analyzing the existing Web and extracting the principles that made it work. This set of principles was written down in the Fielding dissertation which lays out the set of constraints that, when enforced, will make a generic network system into a resilient network like the Web. In Chapter 5 of the dissertation Fielding outlines REST’s interface constraints.

REST is defined by four interface constraints: identification of resources; manipulation of resources using representations; self-descriptive messages; and, hypermedia as the engine of application state.

Why is this important for APIs?

API design is in an infancy. Each API is designed for a single use case and standard. This proliferation of different design ideas results in APIs that each have their own specification and semantics. Interoperability between APIs is nonexistent. In the early 1990s the Web was facing this exact problem and as a result the principles of REST were formalized and adopted. The result was a scalable, resilient and ultimately successful system. By adopting the REST constraints in our APIs we can take advantage of this foundational work to provide APIs that form a scalable and resilient network of their own.

We will take a look at each of the four interface constraints in turn and see how they can be used to design an API that developers love to use.

1. Identification of Resources

A resource is the key abstraction of REST. Anything that can be named is a resource – a video, a document, an image. We identify resources using a unique identifier called the Uniform Resource Identifier (URI). When discussing the Web we almost always use a more specific form of URI called a Uniform Resource Locator (URL). A URL is what we are all accustomed to when accessing sites through HTTP.

By building REST APIs over HTTP we can use URLs to identify resources accessed by our API. To satisfy the REST constraint each URL should map to a single resource and all access to this resource is done through that URL. As an example, if I want to offer an API for a shipping application I might have a resource representing an order. A URL for the individual order numbered 12345 would have a path including the order number.


This simple example shows how we can create a URL to uniquely represent each resource represented by our API.

2. Manipulation of Resources Using Representations

When making a request for a resource the server responds with a representation of the resource. This representation captures the current state of the resource in a format that the client can understand and manipulate. Abstractly the representation is a sequence of bytes along with metadata describing those bytes. This metadata is known as the media type of the representation. Typical API examples are HTML, JSON, and XML. Because the server sends a representation of the resource it is possible for the client to request a specific representation that fits the client’s needs. For example, a client can ask for the JSON representation of a resource or the XML representation of the resource. The server may provide this representation if it is capable of doing so. This concept is called content negotiation. You can use content negotiation in your API to allow multiple clients to access a different representations of the resource from the same URL.

The way that a client asks for a specific representation is through the HTTP Accepts header. The following request is asking for a plain text representation of the order.

GET /orders/12345
Accept: text/plain

Whereas the following request is asking for a JSON representation.

GET /orders/12345
Accept: application/json

By using content negotiation in your API you can offer new resource representations without changing the resources URL or breaking existing clients. This keeps your client and server flexibly decoupled.

3. Self-Descriptive Messages

The representations served by a RESTful system contain all of the data required by the client to understand and act on the resource. If any additional information is needed but not contained in the response a link to that information should be provided within the response. This means that the media type you choose to respond with should be self documenting and should list any related resources or actions that the client may be interested in. Specific media types already exist for JSON APIS. These include HAL, JSON-LD, Collection+JSON and JSONAPI. For a detailed discussion see my post on choosing a hypermedia type

Each of these formats are aiming to solve the same problem:

JSON has no built-in support for hyperlinks, which are a fundamental building block on the Web.

By linking together the resources offered by your API you can build self-descriptive APIs.

4. Hypermedia As The Engine Of Application State

Together, the first three REST constraints imply the fourth and most important – hypermedia as the engine of application state. By uniquely identifying resources, using representations to communicate resource state and using self-descriptive messages via media types all application state stays with the client.

By keeping all application state with the client a direct connection between the client and the server is not necessary, allowing the server to scale to serve many clients with minimal resources. This ability to scale has led to the Web’s scalable and resilient nature.

By designing our APIs with the REST constraints we can build scalable and resilient APIs.