When the requirement is to minimize the effect of a hardware failure on your application, you can use the spread placement groups. Any application with a small number of critical instances that should be kept separate from each other and made highly available can make use of this feature. To place the instances in a spread placement group, we need to start the instances at the same time.