Something that's pretty handy with a lot of servers is that they can do tricks with memory similar to what RAID does to hard drives. That is, there are options for redundant or spare memory. Different machines do this in different ways, so you need to check the user manual for specific configuration options. At the risk of extreme oversimplification, what happens is the machine takes a portion of the available memory and holds it separate. The OS doesn't get to use it. If the main RAM fails, gets errors, or so on, then the server will switch over to that RAM. The downside is that you don't get to use all of your RAM at one time. The upside is system stability and uptime.