| > as the whole server been setup for a very important client of me, and for unplanned downtime, we have to pay huge fine. If your availability requirements are such that you get fined for downtime then you can never ever rely upon a single machine, no matter who hosts it. Even if you have your own dedicated server it will eventually fail. (Be it a dead drive, dead NIC, or blown PSU/PSUs.) You need to be setting up a cluster for high-availability, without a single point of failure. Even then you're at the whims of the network between your clients and the location where you're served - some ISPs will have issues at any given moment, and will have broken routes. Really your problem here seems to boil down to three things: 1. Your server was rebooted and you had no monitoring in place to detect the downtime - pingdom, etc, would have alerted you. 2. You seem to think a single guest/droplet/host will be 100% available. 3. You have a spare host/hosts to failover, but that process is manual, so without advance notice you didn't know to do it. See point 1. |