In 2005, Intel, and later AMD, built extensions to their x86 chip architectures that took much of the
resource-management load off the hypervisors, making it much easier to build a hypervisor that could run guest operating
systems as virtual servers without having to modify them first.
Server operating systems are designed to own the chips, memory and other resources in the machine on which
they run. Without extra functions built into the processor, chipset and BIOS the hypervisor must take responsibility for
marshalling, queuing and executing commands from all the guest operating systems without causing them to conflict, fail or
destabilize the base operating system. Virtualization-enhancing additions from the chipmakers (IVT in Intel’s case) build the
caching and queuing into the hardware itself and provide hooks that both the hypervisor and guest OSes can use to route jobs
directly to the processor without destabilizing themselves or other OSes on the machine. VMware introduced a hypervisor for
x86-based systems in 1999, but the difficulty in resource-management restricted its performance and slowed the progress of
potential competitors until IVT became available in 2005.