Define a successful recovery
Identify the data that matters and whether the application needs preparation for a consistent backup. I would write the expected recovered contents first. A completed backup status does not by itself demonstrate that the application can use the result.
Plan the restored volume
Consider where the restored volume is created and how the operating system and application will recognise it. Permissions and encryption configuration can affect recovery. Write the steps before an emergency rather than discovering them while the service is unavailable.
Use harmless sample data
For practice, restore to a separate target and compare the contents with your expected result. Record the recovery steps and cleanup decisions. Keep important originals protected while learning and follow the retention needs of the exercise.
Try it yourself
Create a recovery test with one sample file, expected contents and a separate restore target. Explain how you will prove the result.
A common question
Is a completed snapshot a completed recovery test?
No. Recovery testing verifies that you can restore and use the data you need.
References & video notes
Use the video to follow the topic with me. Console screens and older offers can change, so check the current AWS documentation before performing a lab.
Keep learning
- Block, file and object storage: choose the access pattern
- Instance store vs EBS: where should your data live?
- EBS volume types: understand IOPS and throughput