Virtualization allows a business to make fashionable "green" claims while alleviating itself of surplus IT employees while lightening the
load on the electric bill, server space, and hardware costs. With virtualization, entire data centers can be consolidated as multiple servers and can run on one server blade with no discernible impact on the delivery of data and applications to end users throughout the enterprise.
Businesses of every size should consider virtualization, but it may not make sense for every server. Before you jump on the bandwagon, you might want to get some more information because there are some applications that are not good candidates for virtualization that can make it a bad technology to adopt.
Image Credit: Wikimedia Commons/Cqq
Measure before You Move
Executives may be sold on the benefits of virtualization, but making a move just because a technology is popular may not be wise. Data needs to be gathered that documents CPU load and memory utilization of every application needs to be measured together with disk I/O and network I/O to determine whether a virtualized environment will benefit the company. Applications such as Novell PlateSpin Recon can map the physical workload of a server and can simplify these measurements. Armed with valid measurements of the performance metrics involved, IT managers can make the case for or against virtualization on a server by server basis.
Some bad candidates for virtualization can be identified by determining the ability of the enterprise to implement and support hardware redundancy. Because of the density of servers per hardware platform, an instance of hardware failure can be disastrous. If the company is not willing to invest in redundant hardware for critical platforms, the company becomes a bad candidate for virtualization: Standalone servers can mitigate the impact of hardware failure.
When measuring CPU and memory usage, IT personnel can identify bad candidates for virtualization. Large databases and other applications can load the host server to the extent where they slow down other servers running on the same platform. When measurements indicate that this is a problem, managers may be tempted to run the server by itself on a host. Unless there is some compelling management or consistency requirement, such a condition would entirely relieve that particular application of any virtualization requirement.
Finalizing the Report
When the software and hardware demands of servers are measured and evaluated, bad candidates for virtualization are identified and a realistic report can be made. Information such as the number of workloads that are good candidates for virtualization to be moved and the number of virtual servers per host can combine with estimates on the percentage of servers eliminated and of the percentage saved on power costs.
Virtualization is a leading technology and virtualization trends will continue to bring tremendous benefits to the table. Still, without identifying bad candidates for virtualization, a company can actually lose efficiency as a result of the investment in technology.
Companies that are bad candidates for virtualization as a whole (i.e. a high concentration of servers that should not be virtualized) can avoid wasting money on technology that won’t help them by spending some extra time on evaluation and planning.