I'm quite surprised no fallback/failover was put in place for a critical tier service like this.