What we did was create an app dedicated to subscribing, with only one worker — we call this our Ingestion System API. It then passes the data to our Process API, which runs with multiple workers for parallel processing. Hope this helps.