We want to move image storage and VM directories to a secondary EBS volume to improve macOS runner IOPs. ## Context We will mount a secondary volume, which, according to AWS, comes with immediate full performance (IOPS/throughput). EBS Volumes created from a backing snapshot require prewarming/hydration, and are slower to reach full performance. We have hit a blocker with running our nesting service with a secondary EBS volume. Macs require an application to be granted Full Disk Access, which can only be done in the UI, causing complexity in the image build process. We have some options on how to automate this process. Now that we have managed to solve the slow ECR pulldown of job images with an S3 method, we can see that the performance issues we observed with the local disk were likely due to copying from prewarming/hydration, not the operations on the local disk itself. However, local SSD sizes are too small to hold all our images and VMs, making them impractical. [Discussion](https://gitlab.com/gitlab-org/ci-cd/shared-runners/infrastructure/-/issues/73#note_1377993519 "SaaS MacOS medium M1 runners Bug Bash Blitz") on slow EBS for macOS runners on AWS. Following [this discussion](https://gitlab.com/gitlab-org/ci-cd/shared-runners/infrastructure/-/issues/73#note_1380807468 "SaaS MacOS medium M1 runners Bug Bash Blitz") from @ajwalker The existing EBS volumes volume setup shows slow IOPs.