The University of Tulsa
With shrinking feature sizes and increasing capacity, memory system reliability is a growing concern. In some datacenter settings, it has been suggested that storage be built entirely out of DRAM memory, and large-scale systems be created by aggregating the main memory of thousands of servers. This places greater pressure on the memory system to not yield errors. High-availability servers are typically expected to provide chipkill-level reliability - the ability to withstand the failure of an entire DRAM chip.