You have the ability to control how your EC2 instances are placed in AWS. You don’t have access to the specific hardware but you can create placement groups to ensure certain criteria is met. For starters you have the cluster placement group, this option allows for all of your EC2 instances to be placed on the same physical rack. This option has a high risk for failure but incredibly fast network speeds. Next up you have spread, in this model you have different EC2 instances on different hardware within the same AZ as well as spread across multiple AZ, you are limited to 7 EC2 instances per AZ. This option is useful if you have critical applications and the risk needs to be minimized however you can’t have many instances with this option. Finally there’s partition and it’s sort of a combination, you have a group of instances placed on a rack of hardware and multiple racks spread across multiple AZ. With this option you can have 100s of instances compared to spread. In short, cluster involves having EC2 instances placed on one rack of hardware, low latency, high risk. Spread involves having each EC2 instance on a different physical server and instances are spread across AZ as well, 7 per. Partition places a group of instances on a partition which is a rack of hardware and having multiple partitions per AZ. It is important to find the placement group that best fits your needs.