When you deploy a fleet of instances that are processing transactions in a distributed manner, you sometimes need to control the placement of the instances in the underlying architecture to allow for certain specific application requirements. You can use placement groups to control this behavior.