By now most people have heard about the different kinds of virtualization out there.  Moving your organization to any virtual environment requires altering how you support and manage your infrastructure and your users.  Where you once had many individual resources each managing their own load, now you have few consolidated resources each sharing multiple workloads. 

In the old model, if you lost a resource you might impact one application or one user.  Now losing one resource could mean many people and many applications are affected.  Your backup and disaster recovery strategy for each individual server might not be sufficient for this new model, as the slightest problem can have a much farther reaching impact.  Each resource must now be given more priority and be more extensively monitored and backed up.  Similarly, your IT staffing model might be as streamlined and efficient as possible in your current environment, but might be lacking necessary skills to properly set up and maintain a virtual environment.  Virtualization is not just a project or a small change; it is a complete shift not only in hardware and software, but in thinking as well.

Shifting Your Infrastructure
When moving to a virtual environment you are taking a distributed set of resources and centralizing them.  To understand fully what that means, let’s consider a few simplified examples of the different types of virtualization. 

Server Virtualization. Instead of a dedicated set of hardware for every server, you use a single, more powerful server to run multiple servers virtually.  You consolidate several distributed devices down into fewer centralized devices.

Desktop Virtualization. Similar to server virtualization, only with your end users’ PCs.  You run multiple virtual desktops on servers and your end users connect in to them over the network.  All of their applications run in the virtual desktop on your server, so your end users can connect via very lightweight and inexpensive devices such as thin clients or mobile devices.  All of the computing power and real work is done in your data center.

To make things even more interesting, all of these can be combined.  Your application server can be running as a virtual server, streaming applications out to virtual desktops, which are all sharing data that lives on a SAN.

In the distributed infrastructure, you would purchase hardware that is powerful enough to run one application at peak usage levels, and dedicated storage for that one application and its backups.  There isn’t a great deal of planning that goes into it, and most application vendors provide hardware requirements that you laugh at as you buy something more powerful.  Planning a virtual environment requires forethought and analysis to ensure all virtual machines or applications that you intend to run on that one physical server have enough resources to run as efficiently as, or better than, they do on their dedicated hardware.  You need to think about total disk space for all virtual machines instead of just one so your SAN is sized correctly, but more than that, you need to intelligently divide up your SAN disk arrays to maximize disk throughput – keeping really disk-intensive applications separated so they don’t impact each other.  You also need to plan a more robust backup environment that can handle the additional load of running multiple virtual servers, as the loss of one physical device now means you lose several servers, applications or desktops.

Likewise, you may buy whatever PC is available at the time when doing a PC refresh or adding a position.  If that PC has a dual core processor and 2G of RAM with an 80G HDD, that is what the user gets and it is generally sufficient for anything they might need to do.  Your IT team has been deploying desktops for a long time, so not a lot of thought needs to go into this, you really just pull the trigger, receive the shipment and begin setting it up.  In a virtual desktop environment you have to plan for resources to be distributed across many different types of users that have different needs and different usage patterns.  You have to consider the people who fill up their 80G HDD and the ones who have 40 application windows open at the same time.  Your environment needs to be sized to handle peak loads so your users can stay as productive as they were on dedicated hardware, and you need to analyze the types of users you have so they can be evenly distributed and not impact one another. 

This is a shift for the IT department, as it requires a major shift in thinking about resources, storage and monitoring. 

Shifting Your Thinking
Your IT department needs to understand the entire business, all of the users and all of the needs for every application.  They need to understand security, business continuity and high availability.  They need to be prepared to shift the time that they spent supporting the distributed resources to managing and monitoring the virtual resources.  Reducing the amount of hardware that can fail also reduces the need to drive around fixing issues, and frees up time to optimize the network, storage and virtual environment.  If your IT staff is not prepared for this change, you could potentially have gaps in support or unproductive IT staff.

Additionally, your IT staff needs a broader understanding of the network that serves up these virtual resources, because network issues, latency or outages can now easily impact a person’s or a department’s entire job function.  Moving from distributed storage to a SAN involves understanding the network that exists between your servers and the SAN.  This could involve learning new protocols and switches, how to optimize traffic for storage, and how to keep your storage network secure and compliant.  With so much more being moved onto a data network each day, you must really focus a great deal of resources on making sure that your network is always available to keep the business running.

Before beginning a major project like this you should analyze the business and the personnel that will be responsible for planning, implementing and supporting of the project.  Limitations in virtualization software must be taken into account and only the virtualization projects that make sense for your business should be considered.  For example, if you have many users that run extensive graphics projects you might not be able to virtualize their desktops just yet.  Likewise if you have an application that is extremely I/O intensive, it might need its own dedicated storage or array.  Only through an in-depth analysis of your business can you determine what issues you may have and plan accordingly.  Lack of IT staff knowledge or experience could result in significant performance issues and ultimately lead to the failure of the entire project and a decrease in end user perception of IT and possibly the entire organization. 

Shifting Your Spending
Is a virtualization project guaranteed to save you money?  No, but if you do it right and understand the full short- and long-term commitments to the project it is very likely.  In the long term you will spend less money on hardware and electricity, but that will be partially offset by the additional money you’ll spend on software licensing and infrastructure upgrades.  Depending on the quality of your current environment and the size of your organization, you could have a pretty hefty initial purchase to make, but when comparing the numbers, you need to take into consideration the fact that you are adding services and functionality, and compare that to the initial costs of providing the same upgrades to your current environment.  You’ll spend less money on physical servers, but more on a new SAN – an added cost, but also an added benefit of shared storage.  You’ll need fewer network switch ports, but faster switches and NICs – an added cost, but a faster and more efficient network.  You need to spend money on monitoring your entire network, but will be able to react quicker to potential issues and abuse.  You might also need to hire a consultant or firm to help you with the planning, but that cost will be much less than if you try to do it in house and the project fails.  If you fully understand the additions and benefits of the project, it is easier to understand this shift in spending and see the long-term savings and potential short-term savings as well.  It also allows you to properly determine the total cost of ownership (TCO) for the project so a proper payback date and ROI can be set.  

Shifting User Perception
Perhaps it is just human nature, but when someone sees a large desktop removed from their desk and a small thin client replacing it they feel as though they have been downgraded.  This perception is understandable, and can either be quickly reversed, or dramatically worsened in the first weeks of the conversion.  Determining your more flexible employees that understand IT and can understand the benefits of the project will help you in rolling out, testing and implementing a virtual desktop or application conversion.  Loop these users in at the start of the project, give them a sense of ownership and roll out to them first.  Listen to their feedback and address their concerns immediately to keep them engaged, and you will very quickly get an understanding of what the rest of the project is going to look like.  You will also work out a lot of your bugs on the more understanding users causing fewer problems when your more difficult users, who are more resistant and ultimately more vocal, go through the conversion.  Offering a speed increase might not be as valuable to some as it is to others, so consider other options to provide additional benefits to those users, such as better remote access or VSS backups of their user profile so they can restore accidentally deleted or overwritten files with a click of a button.  Identify in advance the users who have special needs like the ability to burn DVDs, special printers or smartphones/PDAs and make sure they don’t lose any functionality.  Most importantly, your users need to feel like this project isn’t just another way for the company to save money by taking things away from their employees.  Finding the more vocal employees and getting them engaged and informed early about the global and end user benefits will help spread positive PR.

Shifting Vendor Support and Responsibility
Your vendors may be hesitant to authorize their application running in a virtual environment.  This has much to do with how new virtualization is to the hospitality industry and how easy it is to cause serious problems if you don’t understand the possible issues outlined in this article.  First, you need to fully understand the application’s resource needs and ensure that you won’t be causing significant performance issues by virtualizing it.  Your vendor will be concerned that your users will be unsatisfied with their product, or will call their support team because of slowness that is a result of running virtually.  They also might not fully understand the resource requirements of their application when run virtually because they are used to having dedicated disk, memory and processor.  You will need to demonstrate your ability to provide adequate resources to them, and they might want you to take responsibility for any performance issues initially.  Allow them to put a clause in the contract that performance issues as a result of resource starvation in a virtual environment are your responsibility and not theirs.  This is fair, and if you have done your homework you won’t have anything to worry about anyway.

All of the shifts discussed in this article are intended to point out the sometimes overlooked changes needed when implementing a virtualization project, as well as some potential issues and stumbling blocks you could encounter along the way.  Prior to starting a virtualization project it is critical that you have a full understanding of the total scope, and that you have the right people in charge.  You must be prepared for the inevitable hidden ‘gotchas’ that exist in every project and especially in one that requires as considerable a change in thinking as virtualization does.  Not only will this ease your transition, it will also help you determine the appropriate technologies to adopt, and what might not be the best fit for your organization.  You’ll be better positioned to achieve the maximum return on your investment, and you’ll establish a solid infrastructure that is able to grow with the organization.

Lyle Worthington is the CIO of Horseshoe Bay Resort. He can be contacted at lyle.worthington@hsbresort.com.