Is the partition of the placement group distributed across both ec2 and ebs?

Partition placement groups in Amazon EC2 are designed to distribute EC2 instances across distinct underlying hardware (racks) to reduce correlated failures, but they do not explicitly manage the placement of EBS volumes.

When you use a partition placement group, EC2 ensures that instances in different partitions do not share the same physical racks, which helps protect against hardware-level failures affecting multiple instances simultaneously. This is particularly useful for large distributed and replicated workloads like Hadoop, Cassandra, Kafka, and similar applications.

However, partition placement groups do not provide explicit guarantees about EBS volume placement or availability. EBS volumes have their own availability characteristics that are independent of EC2 placement groups. EBS is designed with built-in redundancy within an Availability Zone, but this is separate from the EC2 instance placement strategy.

For comprehensive availability of both compute and storage, you would need to consider additional strategies beyond just partition placement groups, such as:

1. Using replicated storage solutions at the application level
2. Implementing proper backup strategies for your EBS volumes
3. Considering multi-AZ architectures for critical workloads

In AWS Outposts deployments, partition placement groups can be particularly useful as they allow you to distribute instances across multiple racks within your Outpost, but the same limitation applies regarding EBS volume placement.
Sources
Using partition placement groups for large distributed and replicated workloads in Amazon EC2 | AWS Compute Blog
AWS.Compute.PlacementGroup - AWS Telco Network Builder
Instance placement - AWS Outposts High Availability Design and Architecture Considerations
Optimize Amazon EC2 for AWS Outposts - AWS Outposts