Keeping a web server up 24x7 takes a bit of work. There are security patches that need applied, hackers that you need to keep out and random operating system bugs that can make the system unresponsive. As a someone who has maintained servers for a decade-and-a-half, this is all just part of the job.
But what happens when the facility that hosts the server has issues? Power failures, network backbone connectivity issues and scheduled downtime are events that are out of my control. How do we deal with these problems?
At E Space Communications, we use pairs of servers in different geographic area with separate backbone connections to both distribute the workload and also offer automatic failover when hosting provider outages occur. When a server on the east coast becomes unavailable, it automatically drops out of the rotation and the server in the other part of the country takes over. Most users don't even notice anything happened.
We don't put all our eggs in one basket. Neither should you.