To ensure that the features can be deployed smoothly, one at a time, they need to keep backward compatibility. This means that you need to be able to live in an intermediate stage when service A has been deployed, but not service B. Each change in the microservices needs to be as small as possible to minimize risks, and they should be introduced one change at a time.
Why don't we deploy them all simultaneously? Because releasing two microservices at the same time is dangerous. To start with, deployments are not instantaneous, so there will be moments where out-of-date services will either send or receive calls that the system is not prepared to handle. That will create errors that may affect your customers. ...
