Information Technology Reference
In-Depth Information
within a virtual PC hypervisor. Newer forms of Microsoft's Windows
platform include the Windows on Windows (WoW) virtual hypervisor
for backwards compatibility in legacy applications. Use of a PC hypervi-
sor such as the popular VMWare service can allow a single workstation
to support as many different virtualized desktop configurations as system
resources allow, together with options for capturing point-in-time snap-
shots for automatic rollback during testing and development. This can
greatly reduce design time by eliminating the need to completely reload
a test system each time a setting or application function is altered and by
providing developers with multiple versions of operating systems, brows-
ers, and application software suites with which to test the new application
under development, without the need for individual physical hardware
systems for each.
Virtualized Servers
Virtualization of multiple physical servers onto more powerful central-
ized hosts within the data center can provide significant cost savings in
terms of hardware, energy, and cooling requirements. Since many serv-
ers can operate at a reduced level during off-peak hours, consolidation
onto a smaller number of hosting machines can reduce the power con-
sumed by idling systems. Each virtualized server is merely a collection
of files, rather than a dedicated hardware platform with storage, CPU,
and memory of its own, allowing more efficient utilization of available
resources as well as portability across hardware hosts, as shown in Fig-
ure 10.1.
Automatic load balancing across hosting farms can ensure that
spikes in demand are met with adequate resource allocation to individ-
ual virtualized servers. Resource management requires strict policies
and attention to operational thresholds in order to protect against “vir-
tual sprawl” in the data center and alignment of dedicated resources
within each host to the needs of its virtual machines. Because vir-
tualization allows separation of hardware and software refresh cycles
(discussed in Chapter 11), policies must also be put in place to handle
software update and retirement of virtual as well as physical systems in
order to properly secure the network and to limit complexity in sup-
port requirements.
