← Documentation
AWS
AWS Resource Parking - Supported Capabilities by Service
Exactly what Park Your Cloud (PYC) automatically starts and stops ("parks") across your AWS account, and what falls outside PYC's current scope. This reflects what PYC's resource discovery and parking automation actually does today, not a general product description.
SUMMARY
| Service | PYC Supports | PYC Does Not Support |
|---|---|---|
| EC2 | Start/stop of individual instances | Instances managed by an Auto Scaling Group (park via ASG instead) |
| Auto Scaling Groups | Park/unpark by suspending launches and scaling to 0 | Warm pool instances (billed and controlled separately) |
| RDS / Aurora | Start/stop of RDS instances and Aurora clusters | Resources in deleting/deleted/failed/incompatible states |
| ECS | Per-service parking (desiredCount to 0 on stop; scales to user-supplied count on start). Handles Application Auto Scaling as well | Services using DAEMON scheduling strategy |
| EKS | Deployments and StatefulSets (replicas to 0 on stop; scales to user-supplied count on start). It pauses attached KEDA ScaledObjects | Cluster itself, node groups/worker nodes. Workloads scaled only by a bare HPA (without KEDA) |
BY SERVICE
1
Amazon EC2
Supported
■ PYC starts and stops individual EC2 instances (parking = stop, unparking = start).
■ If an instance is mid-transition (pending, stopping, shutting-down) when a parking action runs, PYC waits and re-checks automatically before giving up.
Not Supported
■ Instances that have been terminated are no longer tracked - PYC marks them as removed rather than acting on them.
■ Instances that belong to an Auto Scaling Group are parked through the ASG (read below), not as standalone EC2 instances.
2
Auto Scaling Groups (ASG)
Supported
■ Parking suspends the ASG's Launch process and sets MinSize and DesiredCapacity to 0, so the group cannot relaunch instances while parked.
■ Any scale-in protection on running instances is cleared first, so they can actually scale down.
■ Unparking restores MinSize and DesiredCapacity (and MaxSize, if supplied) and resumes the Launch process.
Not Supported
■ The Terminate process is deliberately left active - parking an ASG terminates its instances so billing actually stops, rather than just stopping them.
■ Warm pool instances, if configured on the group, are billed and managed separately and are not stopped by PYC's parking action.
3
Amazon RDS Instances and Aurora Clusters
Supported
■ Start/stop for standalone RDS instances.
■ Start/stop for Aurora clusters (handled as clusters, not per-instance).
■ PYC waits through starting/stopping state transitions before reporting failure.
Not Supported
■ Instances or clusters in deleting, deleted, failed, or incompatible states cannot be started or stopped - no action is taken, and deleted resources are marked as removed.
4
Amazon ECS
Supported
■ ECS service parking: parking sets the service's desired task count to 0; unparking restores the desired count specified by the user at the time they schedule the start action.
■ If the service has an Application Auto Scaling target attached, PYC suspends dynamic and scheduled scaling while parked and resumes it on unpark (optionally restoring min/max capacity).
Not Supported
■ Services using the DAEMON scheduling strategy are not supported for parking. PYC excludes DAEMON services entirely, both from resource discovery and parking actions.
Note
The underlying ASG for ECS node-groups could be managed as an Auto Scaling Group. It will appear as an ASG resource in PYC.
5
Amazon EKS
Supported
■ Parking operates at the workload level: Deployments and StatefulSets.
■ Parking scales the workload's replicas to 0; unparking restores the replica count specified by the user at the time they schedule the start action.
■ If a workload is managed by a KEDA ScaledObject, PYC pauses KEDA scaling while parked (via the paused-replicas annotation) and resumes it on unpark, optionally restoring the min/max replica count.
Not Supported
■ The EKS cluster itself is not a parkable resource - PYC manages workloads running on the cluster, not the cluster.
■ Workloads scaled only by a plain Kubernetes HPA (Horizontal Pod Autoscaler) with no KEDA ScaledObject are not supported and are excluded from both discovery and parking.
Note
The underlying ASG for EKS node-groups could be managed as an Auto Scaling Group. It will appear as an ASG resource in PYC.
Note
This page currently covers EC2, ASG, RDS/Aurora, ECS, and EKS - the AWS services PYC's parking automation and resource discovery support today. Other AWS services are out of scope until added to the platform.