blog.storming

At surface level it looks to set an expectation of two things for their hypervisor: an assumed amount of memory allocated to a VM, and the role of the VM (server or client based). Unfortunately they are basing this off of today's model. Example: Server role: paying $6990 per 96GB RAM sounds like 24-32 VMs per box (2cpu,4core) Now when you have a (2cpu,12core) host with 192GB RAM, they lost an additional $6990 revenue and you gained 24-32 more VMs. The new licening model is a way for VMware to get that money back before the cores/RAM grow too quickly.  That is the smart move on their part to capitalize on Intel and AMD's progress. Unfortunately, it stiffles innovation: ie allocating large RAM amounts to VMs for system cache (thinking Citrix PVS and other RAM caching software) that would dramatically reduce SAN IOPS. How thoughtful of them considering EMC is their parent company, since we all can agree that the most compelling reason to increase the RAM on a VM is to decre