-
Notifications
You must be signed in to change notification settings - Fork 152
Home
Welcome!
This is the REST API Design Guide of the National Bank of Belgium. This guide describes all design choices made for the creation of REST-like APIs at the National Bank of Belgium.
Choices choices choices...
There are many ways to implement RESTful APIs and, at this point in time, no clearly winning industry standard. There are many design choices to make:
- architecture: level 2? hypermedia?
- high level design: pagination? filtering? search? sorting? ...
- detailed design: page=2? offset=10? version in the URL or in a header? ...
When there are many options to choose from, development teams can argue forever about the best solution for each use case.
There are various standardization efforts that cover a lot of ground and multiple ways to formalize "REST" APIs (e.g., json-api, OData, OpenAPI, ...) but none that catered for all aspects we wanted to cover in the APIs of the National Bank of Belgium.
While preparing for the development of our APIs, we've decided to create our own design guide, with our own choices, covering our current and known future needs.
Leonard Richardson has listed different "maturity levels" for REST APIs. This is covered in details here: http://martinfowler.com/articles/richardsonMaturityModel.html
The Level 3, "Hypermedia Controls" introduces the notion of "discoverability" of the REST API, allowing your clients to "discover" the possible interactions within the responses provided by the server. This is often referred to as "HATEOAS" (Hypermedia As The Engine of Application State).
Roy Fielding (creator of REST) stated clearly that level 3 of the maturity model is actually mandatory for an API to be called a REST API: http://roy.gbiv.com/untangled/2008/rest-apis-must-be-hypertext-driven
Nevertheless, so far, we have chosen not to implement HATEOAS in our design; although this doesn't mean that we cannot add support for it at a later time. Take this choice with a grain of salt, it remains arbitrary. You might prefer implementing HATEOAS or similar solutions to go further with decoupling your client and server implementations and make your API truly RESTful.
There are tools and libraries that can help you with implementing support for level 3 in an easy way (e.g., Spring HATEOAS, ...).
So to be clear, this is not a truly RESTful API design guide.
If you're curious, we had a very interesting discussion over this here:
- https://github.com/NationalBankBelgium/REST-API-Design-Guide/issues/8
- https://github.com/NationalBankBelgium/REST-API-Design-Guide/issues/9
Our REST API Design Guide covers...
- naming conventions and best practices (e.g., resource naming rules)
- the design of all major elements you would need for business applications
- CRUD
- pagination & associated metadata
- sorting
- filtering
- bulk operations
- concurrency control
- error handling
- ...
Rules and guidelines in this design guide have different levels. Some are optional while some others are mandatory. Also, there are recommendations against some approaches, as well as things that are simply forbidden.
Here's how each term should be understood:
- MUST: it's mandatory (i.e., hard rule)
- MUST NOT: it's forbidden (i.e., hard rule)
- SHOULD: it's heavily recommended (i.e., soft rule)
- SHOULD NOT: there may exist valid reasons not to respect this but the implications should be well understood
- MAY/OPTIONAL: it's optional (i.e., advice)
Take a look at RFC 2119 for more details.
Contributions are welcome. New designs or change proposals should be discussed via issues on this repository.
This project is distributed under the terms of the EUPL FOSS license
REST Resources Design Workflow
REST Resources Single items and collections
REST Resources Many to many Relations
REST Resources Relations expansion
HTTP Status Codes Success (2xx)
HTTP Status Codes Redirection (3xx)
HTTP Status Codes Client Error (4xx)
HTTP Status Codes Server Error (5xx)
Pagination Out of range/bounds
Long-running Operations Example
Concurrency vs Delete operation
Caching and conditional requests About
Caching and conditional requests Rules
Caching and conditional requests HTTP headers
Error handling Example with a single error
Error handling Example with multiple errors
Error handling Example with parameters
Error handling Example with additional metadata
Bulk operations HTTP status codes
Bulk operations Resources naming convention
Bulk operations Creation example
Bulk operations Update example
Bulk operations Create and update example
File upload Simple file upload
File upload Simple file upload example
File upload Complex file upload
File upload Complex file upload example
REST Security General recommendations
REST Security Insecure direct object references