I've being tinkering with a possible architecture for a location based mobile application. The mobile application provides its geolocation, the backend will use multiple 3rd party APIs to fetch data related to this location and return it back to the mobile app as consumer of its topic.

App -> Produces Geolocation -> Backend Consumes and Fetches Data -> App consumes data from backend.

However, the backend will only fetch from the 3rd party apis if the location data doesn't exist already. Since the location data is common for users in that particular geolocation and is also to avoid any 3rd party API requests as much as possible.

In this case, the mobile app(s) are the consumers of the location data, which isn't ideal since it would required large partitions and wouldn't scale very well.

So I believe I would need some kind of reverse proxy, or gateway in front of Kafka to handle all requests, preferably over Websockets.

Below is a high level overview of what I was thinking.

What is the best solution to avoid the mobile app being a consumer and reduce the number of partitions/consumer groups?