[jcromartie]

If you have backups and repeatable deployments, it's not any harder to just spin up another host than it is to replace one of many points of failures in a decentralized architecture.