Obtaining maximum return on virtualization
With all virtualization solutions, there will be an upfront cost to make the jump. With some solutions, especially those that require high-end iSCSI or Fibre Channel SAN storage, that upfront cost can be substantial. Thus, obtaining the best return on your virtualization dollar is important for validating the purchase.
The old adage "never trust an ROI that you didn't create yourself" still holds true today. But one very major difference between typical software ROI numbers and those associated with virtualization is that most virtualization ROI metrics are based on hard dollar return. Take a look back at the dollar calculations in Chapter 1. The typical metrics for RoV expenditures relate the cost of software and hardware to the return on reduced power and cooling costs, reduced provisioning costs, and reduction in overall data center footprint. Each of these is a hard, quantifiable metric.
That being said, the average payback for an investment in virtualization is around 6 months. This metric means that an investment in virtualization will provide positive benefit back to the business within a short 6-month period. In this section, we'll take a look at some of the ways an organization can maximize that payback, specifically associated with hardware, infrastructure improvements, power and cooling issues, and the virtualization software itself.
First and foremost, virtualization's drive towards consolidation will naturally drive an organization towards larger and more powerful server hardware. Additionally, the aggregation of multiple services onto a single host will further drive higher-quality components that incorporate higher levels of redundancy. Thus, the servers of yesteryear are not likely to be well-suited for incorporation into a production virtualization environment.
Or maybe that's just what the consultants want you to hear. One major skill required for a successful virtualization deployment is a deep understanding of systems performance management. An administrator who has a deep understanding of the required and available performance of the systems in the environment will be well-suited for managing that environment with only as many resources as absolutely necessary.
Existing hardware should be measured prior to a production deployment to validate the quantity of virtual machines that can potentially run on this hardware. If this metric is acceptable to the business, it is likely more cost-effective to retain existing hardware for use within the virtualization environment until that hardware's effective end-of-life rather than an immediate purchase.
As virtual machines can be easily, and in some cases uninterruptedly, relocated from physical hardware to physical hardware, the upgrade of a piece of physical hardware is usually a trivial event. Virtual machines are moved from the virtualization host. That host is then powered down and replaced with a new piece of hardware. Virtual machines are then relocated back to the new hardware.
Depending on the virtualization architecture selected, licensing costs may be per processor, per socket, per hosted virtual machine, or combinations of these. That licensing model can impact the decision to keep existing hardware or purchase new. Either way, virtualization licenses for many products typically easily can be relocated from host to host. Thus, the need for an initial hardware purchase to support the environment can be easily delayed. This has the tendency to reduce the early costs for the move to virtualization.
Depending on the virtualization product chosen, the introduction of virtualization into the computing environment can impact areas of its infrastructure. Network ports are one example. If an environment requires 50 servers to run the business and each server utilizes dual-redundant connections to the network, a total of 100 network ports are required plus the additional ports to interconnect the necessary switches and routers.
Making the move to virtualization, the total number of network ports can be significantly reduced. Let's assume that the virtualization solution itself requires only dual-redundant ports for a physical host. Assuming a 10:1 consolidation ratio is realized, the total number of ports can be reduced from 100 to merely 10 to support the 5 physical hosts running the virtual machines.
|This benefit is realized in even greater ways when one understands that a reduction in port count means a move towards smaller switches and less network infrastructure. These smaller network components are typically much less expensive to purchase and maintain.|
For environments making use of keyboard, video, and mouse (KVM) technology for remote access to server consoles, this reduction in total port count can be even greater. Typical virtualization management consoles have the tendency to eliminate or significantly reduce the need for administrators to perform actions on the physical host's console itself. Individual virtual machine consoles are made available to administrators through a network-based interface. This eliminates the costly doubling of network infrastructure associated with remotely accessing server KVM elements or passing them over the network.
Related to security, there is a requirement in some environments for non-IT individuals to maintain access to the data center. This may be a result of their need to work on server consoles due to a software limitation. Allowing these quasi-trusted individuals into the data center can be a security concern. With most virtualization products, the capability exists to remotely access the console itself over the network. When this occurs, these individuals no longer will need access into the secured data center to perform their daily activities.
Power and cooling
We talked about power and cooling enhancements at length in Chapter 1. But one specific area in which these two elements are especially improved through the application of virtualization is where a wholesale infrastructure expansion is required. The cost to augment existing data center equipment with greater wattage and tons of chilling—or to expand the data center itself—in most cases far out-costs a move to virtualization. Costs for both power and cooling often add little additional value to the business, so any reduction there eliminates a liability upon the business that is difficult to link to assets.
Certain software and licensing costs can be similarly recouped. Specific to the costs for OS licensing, Microsoft has announced a bonus for licenses used within virtualization environments. Specifically for certain versions of its Windows Server OS, namely Enterprise Edition and Datacenter Edition, the purchase of a single physical server license bestows additional virtual server licenses for no added cost. The rules are as follows:
- Windows Server Enterprise Edition Each Enterprise Edition server license for a physical server instance also bestows four additional licenses to be installed on virtual server instances. Thus, in a virtualization environment, the purchase of a single Enterprise Edition license nets a total of five total Windows Server licenses.
- Windows Server Datacenter Edition Each Datacenter Edition server license for a physical server instance bestows unlimited licenses to be installed on virtual server instances. Thus, in a virtualization environment, the purchase of a single Datacenter Edition license nets an unlimited number of Windows Server licenses.
|There are two important notes about these rules: First, the purchase of a single physical license only bestows the additional virtual licenses onto a single server. Thus, the unlimited virtual licenses gained through the purchase of Datacenter Edition only licenses those unlimited virtual instances onto a single server. Second, be aware that the licensing rules state that this limitation is for running instances. You can host as many non-running instances as you want. That being said, if more non-running instances are present on the system, in order to fulfill your licensing requirement, a running instance must be powered off in order to power on a non-running instance.|
Different virtualization architectures have differing capabilities of hosting this number of virtual machines simultaneously. Due to the sharing benefits of OS virtualization, there is a higher potential for server density than with hardware virtualization. Thus for any particular set of hardware, when using an OS virtualization product, there is a higher likelihood of financial gain associated with an upgrade to Datacenter Edition than with other architectures.
Best practices in implementing virtualization
Virtual environments are different than physical environments
Potential usage scenarios and best practices
Obtaining maximum return on virtualization
Best practices in systems automation
About the author: Greg Shields is an independent writer, speaker and IT consultant based in Denver. With more than 10 years of experience in information technology, Greg has developed extensive experience in systems administration, engineering and architecture, specializing in Microsoft, Citrix and VMware technologies. He is a contributing editor for both Redmond magazine and Microsoft Certified Professional magazine, authoring two regular columns along with numerous feature articles, webcasts and white papers. He is also a highly sought-after instructor and speaker, teaching system and network troubleshooting curricula for TechMentor Events, a twice-annual IT conference, and producing computer-based training curriculum for CBT Nuggets on numerous topics. Greg is a triple Microsoft Certified Systems Engineer (MCSE) with security specialization and a Certified Citrix Enterprise Administrator (CCEA). He is also the leader of the Realtime Windows Server Community.