We don't. We don't care about instances at all. We have pipelines that bake base amis. Another pipeline builds the code from a repo and turns it into an artifac

We don't. We don't care about instances at all. We have pipelines that bake base amis. Another pipeline builds the code from a repo and turns it into an artifact. Then the artifact is baked into a new ami from the base ami. Then these are launched and destroyed inside asgs at will. If there needs to be persistant storage this is normally backed by an ebs volume which can then be backed up to other volumes or services like s3. Databases are backed up by another job in a similar matter on a set schedule.