Generic applications, such as web apps or mobile app backends, usually involve calls to remote services and databases. For such cases, it can be useful to take advantage of asynchronous frameworks, such as the ones presented in Chapter 6, Implementing Concurrency; this will improve application logic, system design, responsiveness and, also, it will simplify the handling of network failures.

Use of asynchronous programming also makes it easier to implement and use microservices. A microservice, although there is no standard definition, can be thought of as a remote service that focuses on a specific aspect of the application (for example, authentication).

The idea behind microservices is that you can build an application by composing different microservices that communicate through a simple protocol (such as gRPC, REST calls, or through a dedicated message queue). This architecture is in contrast with a monolithic application where all the services are handled by the same Python process.

Advantages of microservices include strong decoupling of different parts of the application. Small, simple services can be implemented and maintained by different teams as well as be updated and deployed at different times. This also allows microservices to be easily replicated so that they can handle more users. Additionally, since the communication is done through a simple protocol, microservices can be implemented in a different language that can be more appropriate than Python for the specific application.

If the performance of a service is not satisfactory, the application can often be executed on a different Python interpreter, such as PyPy (provided that all the third-party extensions are compatible) to achieve sufficient speed gains. Otherwise, algorithmic strategies as well as porting bottlenecks to Cython is generally sufficient to achieve satisfactory performance.