Asynchronous versus synchronous

Switching to an asynchronous model means you will need to use asynchronous code all the way down.

For example, if your microservice uses a Requests library that is not asynchronous, every call made to query an HTTP endpoint will block the event loop, and you will not benefit from asynchronicity.

And making an existing project asynchronous is not an easy task because it changes the design completely. Most projects that want to support asynchronous calls are redesigning everything from scratch.

Note

The good news is that there are more and more asynchronous libraries available, which can be used to build a microservice. On PyPI, you can search for aio or asyncio. This wiki page (https://github.com/python/asyncio/wiki/ThirdParty) is also a good place to look at.

Here's a short list of those that are relevant to building microservices:

• aiohttp.Client: ;Can replace the requests package
• aiopg: ;PostgreSQL driver on top of Psycopg
• aiobotocore: ;AWS client--might be merged...