I would advise that you keep at least one node or even part of the cluster into a different datacenter thus allowing for a part of your data to be "alive" as such somewhere else.
The backup is as good as the time taken (assuming restore procedures are in place and tested). During the time that has lapsed since the backup your data are not safe unless you are able to have them live somewhere else.
In any case if such a catastrophic situation occurs, you assume that you keep the backups "elsewhere" in order to restore them ?
If that is the case, then keep a spare node there as well.