Imagine that you need to host an infrequently used service, but paying for a dedicated server that is always up and running is proving to be costly or inefficient to maintain. Serverless architectures might be what you are looking for. The name serverless is a misnomer since all client requests are indeed handled by servers, which are dynamically provisioned for the lifetime of the request.

A more appropriate term would be Function as a Service (FaaS), as these platforms support execution of an application logic like a small Python function but does not store any state. Building an application composed of such functions would be quite similar to the microservices architecture discussed earlier.

Typically, you only pay for the milliseconds of server time that a serverless application uses, which makes it much cheaper than dedicated servers. Scaling is automatically handled, so there is no additional effort needed to handle massive spikes in traffic. Last but not the least, there is no headache of having to set up and maintain server infrastructure.

Django might not sound like it would work in such an environment, but Zappa makes it easy to deploy Django applications (in fact, any WSGI compatible application) on a serverless platform such as AWS Lambda with minimal changes. This opens up the possibility of enjoying all the advantages of serverless while using Django.