Intel® Memory Drive Technology Increases Server VM Capacity
Server VM capacity and the need for more memory. Running a large number of VMs on a single server (by either single or multi-tenant) is an established best practice for increasing the utilization of servers. The ability to maximize the number of VMs per server has a direct inﬂuence on the economics of an infrastructure. With the increase in core count and CPU capabilities, capacity planning in VM deployments is now more focused on memory requirements, as the system could run out of the memory resource without maximizing the computing power of the CPUs. This storage capacity limitation has driven the use of less powerful CPUs and curb the memory of workloads running VMs. This use of insufficient amount of physical memory results in the host using Linux* Swap (even with high-end SSDs), typically producing poor performance. In most cases, customers consider this option as a last resort to prevent out-of-memory crash, and carefully plan capacity to avoid using Linux* Swap.