I have a server for a client in our colo, that has been up for 1753 days. (Client doesn't want to move to new hardware, etc. otherwise we would have moved them

I have a server for a client in our colo, that has been up for 1753 days. (Client doesn't want to move to new hardware, etc. otherwise we would have moved them long ago). During that time, there was maybe network outage (at 2AM mostly) of perhaps 2 hours. Complexity reduces reliability. Claiming AWS' uptime approaches that of fairly standard colocation in a decent facility, is laughable.