VMware’s memory management technologies (collectively called memory overcommit) has included these features: transparent page sharing, balloon driver, and hypervisor swapping. With vSphere 4.1, a new category is introduced in the form of memory compression.
Memory compression is an incremental improvement in the most constrained resource in virtual environments today; it is not a cure-all for memory intensive virtual workloads. In the course of my virtualization journey, I’ve added memory to hosts more frequently than processors (in the form of an additional host) or network connectivity; this is usually due to workloads restating their requirements midway through their lifecycle. (As a general rule, workloads are allowed to restate the memory requirements, and I’d rather deal with that instead of a large initial overprovisioning.)
Read the entire post in the Servers and Storage blog at TechRepublic.
Rickatron Blog 