We recently upgraded an MSSQL RDS database instance from 12.00.6329.1.v1 to 15.00.4236.7.v1. The instance functions fine, but it is supporting a legacy application that sometimes encounters errors that leave us no choice but to rollback to a snapshot. We also copy snapshots to our QA account periodically to refresh our QA databases, and this procedure has increased by the same significant amount of time. Prior to upgrading, these snapshot restores took around 30 minutes to get the instance to the Available state. After upgrading, it is taking several hours. The most recent attempt took 6 hours. This is harmful to our business as the increase in recovery time causes cascading effects. This feels like a possible AWS bug, I'm not sure what we could possibly do differently to improve this outcome. Any ideas?
Instance details:
Instance class: db.r6i.4xlarge
Storage type: gp2
Storage size: 2000GiB