Following this morning's incident, we'd like to know how to best set up a fallback db. Does anyone have a solution they like?