Asynchronous messages

In microservice architecture, asynchronous calls play a fundamental role when a process that is used to be performed in a single application now implicates several microservices. We touched briefly on this in the previous chapter with our change to the Jeeves application, which now communicates with its workers using an asynchronous message queue. To make the best use of these, we will investigate these tools in more depth.

Asynchronous calls can be as simple as a separate thread or process within a microservice app that is receiving some work to be done, and performs it without interfering with the HTTP request/response round trips that are happening at the same time.

But doing everything directly from the same Python process is not very robust. What happens if the process crashes and gets restarted? How do we scale background tasks if they are built like that?

It's much more reliable to send a message that gets picked by another program...