We'd like you to develop a basic feature flagging API. The API should consist of a single endpoint to return a list of flags for a given environment. The models you should need are Feature, Environment and Flag. The Feature and Environment models should just have a name (and any other metadata you may deem necessary). The Flag model should have a relationship with the Feature and Environment, as well as a boolean value to describe whether the feature is enabled or disabled in a given environment. The API endpoint should return the enabled / disabled state of the feature and the name (+ metadata) for the feature. You need not return any information about the environment.
We have created the django application, flags, in the apps directory and added some boilerplate code for you,
including a stub of the function based view we expect you to complete. If you would rather use a class-based view, you
may, but you should justify in your response why you chose to do so.
Some notes / clarifications:
- To determine the environment, you may use a header or a query parameter. You may use an authentication class, but are not required to. It must not be possible, however, to request flags without an environment.
- You need not worry about API endpoints to manage these objects, you can use the django admin as necessary.
- You may use the default sqlite database.
- You should not implement any pagination on the list flags endpoint.
Your code will be judged based on:
- How well your solution meets the requirements
- The readability and structure
- The quality of any tests you choose to write
As an extension, by assuming that you will only ever have a single instance of the API application running, implement an in-memory cache system which determines when flags were last retrieved for an environment. You should NOT use a database field for this. You can surface this information via a new (unauthenticated) endpoint, or in the django admin panel.
As a final extension, you are tasked with deploying this application which must be able to scale from 50 requests per second to bursts of 2,000 requests per second. Your task is to build the platform that can serve the requests to the API.
The priority of requirements are listed as follows with the most important first.
- Uptime/Reliability
- Low Latency
- Simplicity
- Cost
In fewer than 8 paragraphs, describe how you would achieve this. Key things to think about are: Overall design, platform choices, any additional tools, management & maintenance.
NOTE: this is a written exercise, you do not need to write any code for this.