feature request: Elasticache Redis autoscaling supports same level of AZ control as manual scaling operations

When ElastiCache auto scaling adds new shards to your cluster mode enabled replication group, the AZ placement of the new shard primaries and replicas is handled automatically by ElastiCache. The service distributes nodes across the availability zones in your subnet group to maintain high availability and fault tolerance.

However, based on the current auto scaling capabilities, there is no way to specify AZ placement constraints for nodes created through auto scaling operations. You cannot direct auto scaling to create a primary in a specific AZ (like us-east-1c) and replicas in other specific zones. The auto scaling feature focuses on adjusting capacity based on metrics like CPU utilization or through scheduled scaling actions, but it does not provide granular control over the availability zone placement of the nodes it creates.

This represents a gap between manual scaling operations, where you have more control over node placement, and auto scaling operations. Your feature request for AZ-level control in auto scaling operations would allow for more precise architectural control while still benefiting from the automation that auto scaling provides. This would be particularly valuable for workloads with specific compliance, latency, or availability requirements tied to particular availability zones.
Sources
Use CloudFormation for Auto Scaling policies - Amazon ElastiCache
ElastiCache and Application Auto Scaling - Application Auto Scaling
Applying a scaling policy - Amazon ElastiCache
Scheduled scaling - Amazon ElastiCache