Memory mirroring is the division of memory on a server into two channels. The first channel is mirrored to the second channel, creating a redundant copy of memory.
Memory mirroring can be compared to RAID 1 for storage. If a fault occurs within the memory of one channel, the memory controller shifts to the paired channel without disruption, and the channels can re-synchronize when repairs are completed. The disadvantage of memory mirroring is also the same as RAID 1 in storage -- memory costs are effectively doubled because memory contents are duplicated.
Increased memory allows for higher levels of consolidation, and the reliability of that memory affects the overall reliability of all the virtual machines (VMs) on that server. As such, memory features include advancements such as fault-tolerant memory mirroring and memory sparing. In sparing mode, the trigger for failover is a preset threshold; when the threshold is reached, the content is copied to its spare and the spare counterpart is activated for use. Mirroring can be used along with memory sparing to allow for the installation of hot spare memory modules, which the server can use if it detects a problem with an active memory module.
See also: database mirroring