ECS Cross-AZ Disaster Recovery (Whole-Instance Image)

1. Scenario

This Skill guides the user through creating a whole-instance image (system disk + all data disks) from an existing ECS instance, and deploying a new instance in a different availability zone within the same region using that image, to achieve AZ-level disaster recovery. The original instance and all its resources remain completely untouched -- this is a backup, not a migration.

Architecture: ECS Source Instance -> ECS Custom Image (full disk mappings) -> ECS New Instance (target zone) + VSwitch (existing or newly created in source VPC)

Key characteristics:

• No downtime: create-image supports Running instances
• Whole-instance image: automatically includes mappings for the system disk and all data disks
• Single-shot creation: use run-instances --data-disk Device=... to override disk Category/PL on the image during instance creation, avoiding separate disk attachment
• All parameters explicit: network / billing / bandwidth / disk PL are read from the source instance and reused

2. Installation