This actually is a great use-case for Lambda functions, since it is not possible to process everything parallel on a single machine, and it would be very hard to manage if you would scale a lot of instances up and down all the time based on processing queues.