The idea of competing consumers is simple: you have an input queue (or a messaging channel) and a few instances of consumers that fetch and process items from the queue concurrently. Each of the consumers can process the message, so they compete with each other to be the receiver.
This way, you get scalability, free load balancing, and resilience. With the addition of the queue, you now also have the queue-based load leveling pattern in place.
This pattern integrates effortlessly with priority queues if you need to shave latency from a request or just want a specific task submitted to your queue to be performed in a more urgent manner.
This pattern can get tricky to use if the ordering is important. The order in which your consumers receive and finish to process messages may vary, so make sure that either this doesn't impact your system, or you find a way to reorder the results later on. If you need to process messages in sequence, you might not be able to use...
