We designed out RabbitMQ model as having one queue per customer.

This way we can guarantee concurrency and FIFO for each of them separately and maximize distribution across our consumers.

However this poses a big question to us, we will have for each consumer, consume from all these queues.

Which for short means, we have to loop through all of our customers, call the channel.consume on their respective queues.

If another customer is added, we will need to add accordingly, maybe do it every one or two minutes, so new queues also getting consumed.

Is this a good idea. Or we will run into consumer side bottlenecks such as long time for setting up consumption or updating these.

Also as the number of customers grows , we have a growing number of consumption plus other supportive queues.

What do you think of our design choice?

We want some feedback on this design of if there is a better way to consume many queues at once. Our idea is similar to sharding, however we are controlling the shards and exactly to which they are going to. The biggest issue is we cannot consume the as one logical queue, which would be perfect to solve this ever increasing number of queues issue.