Because that's an overly complex solution. Why not just fix the server capacity problem (or whatever is causing the outage)? Seems much simpler and they've been able to handle this in past years.