I want to use Kafka as a message broker on the server side of a distributed application. We have clients that can be located around the world, and that send frequent messages to the server. Right now those messages are just JSON over ZMQ. I am considering changing the connection between client and server to Kafka (since I'm going to use Kafka in the server side anyway) but have the following confusion:
Is Kafka even meant to be used this way (distributed clients connecting to the server with it)? I don't see that in the docs.
If I don't use Kafka, I'll have a single choke point at my server/DC entry. What's the efficient alternative to get information from the software gateway (which is now ZMQ) into kafka, without having that software gateway be a bottleneck?