Best Practices For REST API Design

Why This Blog?

This blog highlights REST Best Practices intended for the developers who are interested in creating RESTful web services that provide high reliability and consistency across multiple service suites. We all know there are various resources on best practices for creating RESTful web services.

However, it should also be known that many of the available resources are conflicting. Also, it’s likely not doable to refer, read, and comprehend several books on this subject with a motto of implementing services “tomorrow”. This blog will enable you to quickly understand and grasp RESTful concepts with a major concern on best practices and conventions.

What Is REST?

REST is an abbreviation of Representational State Transfer. REST represents the architectural style of the World Wide Web (WWW). REST is an approach to communication that makes simple HTTP calls for inter-machine communication. REST is often preferred over SOAP because REST does not use as much bandwidth as heavyweight SOAP does which makes it a better option for use over the Internet.

In REST, the calls will be message based and dependent on HTTP standards to describe these messages. REST can be understood as a simple request/response mechanism, where each request returns a response. The following fig. will give you a better idea:

RESTful API’s, more popularly termed as REST API’s are the web services written using REST architecture

REST is resource-based. When we talk about REST, it indicates things or resources as opposed to actions in SOAP. In RESTful services noun represents the resources, for example, a person resource, a useful resource.

We use an HTTP verb (POST, PUT, GET, DELETE, etc.) to indicate what operation the REST API has to perform. Resources are identified by URIs. It is possible to have multiple URIs pointing to the same resource. RESTful API’s generally accept/return data in json or xml format.

Let’s quickly form an example with this basic knowledge before moving further. Suppose Sagar is a resource, there is a service to fetch contact information using the HTTP verb GET which will return the data like name, address, phone number, email address in JSON or XML form.

We Helped Our Client To Convert The Old Application Into A New Stack

REST architectural style defines the following six constraints:

  1. Uniform Interface
  2. Stateless
  3. Cacheable
  4. Client-Server
  5. Layered System
  6. Code on Demand

Let’s have a small review of each:

1. Uniform Interface:

It defines the interface between clients and servers. Uniform Interface simplifies and decouples the architecture. It enables each part to develop or evolve independently. Uniform interface describes the following four guidelines or principles:

  • Resource-based
  • Self-descriptive messages
  • Resource modification using Representations
  • Hypermedia as the Engine of Application State (HATEOAS)

2. Stateless:

It is not required for the webserver to remember the client’s state. The necessary state required to handle the request is included in the request itself, whether as part of URI, query-string parameters, body, or headers.

3. Cacheable:

Clients can cache responses. Therefore Responses must explicitly or implicitly define themselves as cacheable or not which prevent the clients from reusing stale, old or inappropriate data in response to further requests.

4. Client-Server:

The uniform interface is responsible to separate the clients from the servers. Servers and clients can be developed independently. It conveys that the clients are not concerned with data storage which is internal to each server, which increases the portability of the client code. Servers are not concerned with the UI or user state, which makes the servers more scalable and simpler.

5. Layered System:

A client is unable to tell if it is connected directly to the end server or to an intermediary server. Intermediary servers may improve system scalability by providing shared caches and enabling load-balancing. Security policies can also be enforced through layers.

6. Code on Demand:

It is possible for the servers to extend or customize the functionality of a client on a temporary basis by transferring logic to it that it can execute, for example, Java applets and client-side scripts like JavaScript.

HTTP Verbs/ Request Methods:

Clients specify their desired interaction method in the Request-Line part of an HTTTP request message. Each HTTP method is bound to some specific, well-defined rules and meaning within the context of a REST API’s resource model.

The primary or most commonly used HTTP verbs or more properly called methods are GET, PUT, POST, and DELETE. These correspond to create, read, update, and delete operations respectively. OPTIONS and HEAD are the other verbs but are less frequently used.

GET:

The GET method is used to read or retrieve a representation of a resource. As the GET method (along with HEAD) is used for reading only purposes hence they are considered safe as they do not change the data. GET (and HEAD) are considered as idempotent, i.e., making multiple identical requests end up having the same result as a single request.

Examples:

GET http://www.myexample.com/users/orders

GET http://www.myexample.com/users/12345

GET http://www.myexample.com/users

PUT:

We use the PUT method mostly for update functionalities. Use this method for PUT-ing or updating data to a known resource URI along with the request body which contains the newly-updated representation of the original resource that we intend to change.

PUT method can also be used to create a resource in cases where resource ID is provided by the client instead of the server. But we must generally use POST to create new resources where the client-defined ID is included in the body representation.

PUT is not a safe method, although it is idempotent.

Examples:

PUT http://www.myexample.com/users/123

PUT http://www.myexample.com/users/orders/123

POST:

POST method is utilized to create new resources. This method is not safe nor idempotent.
Examples:

POST http://www.myexample/com/users
POST http://www.myexample.com/users/orders/123

DELETE:

As the verb suggests, it is used to delete a resource identified by the URI.

Examples:

DELETE http://www.myexample.com/users/123

DELETE http://www.myexample.com/users/123/orders

Response Status Codes:

Forty standard codes have been defined by HTTP to convey the results of a client’s request. These codes can be divided into the following five categories:

CATEGORYDESCRIPTION
1xx InformationalIndicates transfer protocol-level information.
2xx SuccessIndicates that the client’s request was accepted/processed successfully.
3xx RedirectionInforms that the client needs to take some additional action to complete their request.
4xx Client ErrorThis category of error status codes defines the client’s error.
5xx Server ErrorThese error status codes indicate the server side errors.
All ItemsDescription

Let’s have a short and precise look over the success and error status codes.

HTTP response success codes:

CODENAMEMEANING
200OKIndicates a nonspecific success
201CreatedIndicates that a new resource has been created
202AcceptedIndicates the start of an asynchronous action
204No ContentIndicates that the body was left blank
301Moved PermanentlyIndicates that a new permanent URI has been assigned to the client’s requested resource.
303See OtherSent by controllers to return results that it considers optional
304Not ModifiedSent to preserve bandwidth (with conditional GET)
307Temporary RedirectIndicates that a temporary URI has been assigned to client’s requested resource.

HTTP response error codes:

CODENAMEMEANING
400Bad RequestIndicates a non-specific client error.
401UnauthorizedIndicates either client provided wrong/invalid credentials or missed to send them.
402 ForbiddenIndicates denied access to a protected resource.
404Not FoundSent when the client tried interacting with a URI which the REST API could not map to a resource.
405Method Not AllowedSent when the client tried to communicate via an unsupported HTTP method.
406Not AcceptableSent when the client tried to request data in an unsupportive type format.
409ConflictIndicates that the client attempted to violate resource state.
412Precondition FailedInforms the client that one of its pre-condition was not met.
415Unsupported Media TypeSent when the client submitted data which is of unsupported media type format.
500Internal Sever ErrorTells the client that API is having problems of its own.

Best Practices/Tips:

1. Use Noun instead of verbs:

Nouns are good and verbs are bad when it comes to rest standards. You can use the following easily understandable structure for every resource:

RESOURCEGET (READ)POST (CREATE)PUT (UPDATE)DELETE
/customersReturns a list of customersCreate a new customerBulk update customersDelete all customers
/customers/1234/customers/1234(405 – Method not allowed)Updates a specific customerDelete a specific customer

Bad practice is to use verbs like:

/getAllCustomers

/createNewCustomer

/deleteCustomers

Nouns are good and verbs are bad when it comes to rest standards. You can use the following easily understandable structure for every resource:

2. GET method and query parameters should never be used to update or alter the state:

Never use GET to alter or state changes. Use GET, POST or DELETE to alter the state.

3. Use plural forms of the noun:

Keep the resource naming simple by following plural forms of nouns. Even though you might think that it is grammatically incorrect, but the best standard for REST practices is to use plurals.

/customers instead of /customer
/customers/12345 instead of /customer/12345

4. Use sub-resources for relations:

If you have some resource which is related to another resource then it is implied to use sub-resources:

GET /customers/1/products: Returns a list of products purchased by customer 1

5. Use HTTP headers for serialization formats:

The client-side and server-side both should know and use the format that is being used for communication. This format has to be specified in the HTTP-header.

Accept: It defines a list of acceptable response formats.
Content-Type: It defines the request format.

6. Error handling and using status codes:

It is very difficult to work under situations where proper error handling is ignored. Just returning of HTTP 500 with a stacktrace log is not enough.. We need to make proper usage of the HTTP status codes that have been covered in the earlier part.

7. Include scope for paging, sorting, filtering and field selection:

Pagination:

Use limit and offset for pagination. It is not only flexible for the user but also common in the leading databases.

  • GET /customers?offset=5&limit=20

Sorting:

Ascending and descending sorting should be provided. Also allow sorting over multiple fields.

  • GET /customers?sort=-age,+lastName

Filtering:

We can make use of unique query parameter for all fields or may be use a query language for filtering purpose.

  • GET /customers?age>=24 Returns a list of customers whose age is equal to greater than 24.

Field Selection:

It is a very common practice that mobile clients wish to display only a few attributes. They don’t want to display or use all the attributes of a resource. Thus an API should have the ability that allow the mobile clients to choose the returned fields. This helps to save the mobile battery, reduce network traffic, and speeds up the API usage.

  • GET /customers?fields=name,age,contactNumber

8. Documentation:

I hope the above content is easily understandable and lets you get started with developing REST API with some good level standards and consistency in your coding style. Just follow these REST best practices and you’re all set.

One last suggestion is to maintain an updated API document. This document should be shared and accessible to all concerned people. Most client-side developers should check the API document before starting the integration of web services. The doc should cover all necessary information like request URL, response body format, headers, etc. For every API try to cover with an example of request body and response bodies with all the alternative responses including error and success.

Content Team

This blog is from Mindbowser‘s content team – a group of individuals coming together to create pieces that you may like. If you have feedback, please drop us a message on contact@mindbowser.com

Keep Reading

Keep Reading

  • Service
  • Career
  • Let's create something together!

  • We’re looking for the best. Are you in?