Beruflich Dokumente
Kultur Dokumente
P
ropelled by the technologies that The company implemented VMware®
VMware and Intel have pioneered, Infrastructure 3 featuring VMware® ESX
virtualization has rapidly gained server hosts on HP ProLiant servers pow-
market acceptance. Indeed, VMware ered by quad-core Intel® Xeon® proces-
virtual infrastructure software has some sors. By going from 65 physical servers
4 million users and more than 20,000 to just 17, McMillan says the virtualiza-
corporate customers, many of which have tion implementation “has saved us $1.5
been built on Intel-based systems. million in pure hardware costs. That’s not
One of those corporate customers is even counting power and cooling costs, or
Solvay Pharmaceuticals, a chemical and staffing costs.”
pharmaceutical group with 29,000 em- There are many more examples of suc-
ployees in 50 countries, which identified cessful virtualization deployments across
virtualization as the solution to contain many industries and mixes of applica-
server growth. “Because we’re in a regu- tions, leading to the emergence of clear
lated industry, we would need to get three best practices that can benefit organiza-
new servers every time we needed a new tions implementing virtualization for the
application—one for development, one first time.
for test and one for production,” explains
Bruce McMillan, Solvay’s manager, The Basics
emerging technologies. Let’s start by reviewing some of the
hardware and software elements of virtu-
alization. There are several virtualization
architectures or approaches. For example,
a hosted virtualization architecture runs
the virtualization layer as an application
on top of an operating system. VMware®
Server for Windows or Linux is an example
of a hosted architecture. The virtualization
layer in a hosted architecture is at times
referred to as a monitor.
In contrast, a hypervisor (also called
bare-metal) architecture installs the
virtualization layer directly on a system
without relying on the OS beneath it. With
direct access to the hardware resources,
a hypervisor can be more efficient than
a hosted architecture and offers better
scalability and performance. VMware
response time and application throughput. don’t adapt well to multiple threads, some
• Deployment strategies emphasizing organizations partition their servers into a
two-socket servers had the lowest TCO per number of VMs, each with the appropriate
VM when server demand was relatively low, number of CPUs to match the number of
which can occur as a result of business threads supported in the application.
policy requirements, limitations imposed • Adequately train staff. Virtualization can
by the technology infrastructure and/or be simple, but not all implementations are
limited demand for compute resources at easy. Therefore, it is important that staff
small sites. understand all of the issues associated
Overall, Intel IT’s performance com- with virtualization and allow sufficient time
parison found that four-socket systems to bring up new VMs.
offer better performance per watt and • Implement a cross-functional team to
better TCO. Organizations must look at the oversee virtualization. Just as in a purely
computing load to be virtualized in order to physical environment, undisciplined virtu-
determine whether to proceed with scale- alization can lead to sprawl as more VMs
up with four-socket systems, scale-out with are added. This should be a process that
two-socket systems, or a combination of has structure and discipline to ensure long-
the two. (To read the white paper, “Com- term performance and manageability.
paring Two- and Four-Socket Platforms for • Avoid becoming “network-bound,” which
Server Virtualization,” visit http:// occurs when activities such as VM backup
communities.intel.com/docs/DOC-1512.) become constrained by limited network
capacity. Again, this is an issue that can
Other Considerations be anticipated with proper planning.
• When implementing virtualization, avoid • In a virtualization implementation, I/O
putting all high-CPU demand virtual serv- is typically the most substantial expense.
ers on the same physical server. By imple- In other words, a high level of I/O require-
menting VMware DRS, computing capacity ments will have a greater cost impact.
is dynamically allocated and balanced Although the concepts presented here do
across hardware resources aggregated into not cover every possible situation, they can
logical resource pools. serve as a guide for starting the process of
• If you are planning a virtualization virtualization planning.
implementation with existing servers and VMware and Intel offer combined exper-
are running out of power, connectivity or tise to ensure infrastructure resources are
cooling, consider a hardware refresh. designed, manufactured and optimized
• Deploying newer, higher-performance to handle evolving data center require-
servers that can host more virtual ma- ments. They are continually collaborating to
chines leverages existing virtualization introduce innovative hardware and software
software licenses. enhancements to provide customers with
• Consider depreciation. Normally, in a improved security, reliability and manage-
traditional physical server deployment, the ability. Their close partnership allows users
payback is three to five years. However, to enjoy near-native performance of appli-
with virtualization, savings on power and cations running in virtualized environments.
management indicators can result in pay-
back in less than one year. For more information about VMware, Intel and
• Consider industry- and application-specif- their virtualization solutions, go to www.vmware.
ic issues. For instance, for applications that com/go/intel.