I guess multiple datacenter hosting and automatic switch over should be the resulting solution of this incident