Five Steps to Determine When to Virtualize Your Servers

Server virtualization isn’t just for big companies. Entry-level

virtualization tools are free or low-cost, and there are many

benefits to virtualization (including saving money). It’s not

a question of “if” you should virtualize your servers;it’s a

question of “when.” In this article, I outline five steps you

should take to determine when to virtualize your servers.

There are a number of server virtualization solutions available

today. However, this article isn’t about which solution to

choose. Many virtualization questions are “solution agnostic,”

and the question of “when” to virtualize your servers is one of

them.

So, if you haven’t started using virtualization or you haven’t

fully virtualized your IT environment, I recommend the

following five steps to determine when you should make

that move.

Step #1 – Understand the Benefits of going virtual

You don’t want to undertake a virtualization project without

understanding why you are making this effort. Most of us

have to justify a project like virtualization to a manager,

director, VP or CIO. Even if you don’t have to do that, you

should be able to answer the “why” question for yourself with

an answer that’s more concrete that “because it’s the next

big thing.” Below is a list of reasons why most administrators

feel compelled to virtualize their server infrastructure:

• Save time – Administering virtualized servers over physical

servers can save a huge amount of time.

• Save money – Any way you measure it, virtualization comes

out as a cost-saving proposition. Dollars are saved in less

administrative time, fewer infrastructure requirements and

less energy utilization.

• Simplify management – Virtualization enables the use

of advanced features like resource optimization, high

availability and point-in-time snapshots of servers.

• Recover from disaster – Having a reliable disaster

recovery plan is essential for ensuring business continuity.

Virtualization offers hardware independence and decreased

recovery time in case of a disaster or failover.

Once you and your management team are convinced that

virtualization is the right decision for your company, move

on to the next step.

Step #2 – Evaluate a virtualization solution.

There are many virtualization solutions available today. In

addition to VMware vSphere 4, you can evaluate Microsoft

Hyper-V or Xen/Citrix, to name a few. If you choose to

evaluate or analyze all of them, you may be spinning your

wheels. To date, VMware has held the dominant position

in the virtualization market space, with more than 150,000 customers globally. No one else in the marketplace has come close to matching the maturity, breadth of offerings, reliability, or adoption rate of VMware. Many times, all that the competition can offer is a claim of a lower price tag, but make sure that you’re doing an apples-to-apples cost comparison – more on that later.. When selecting the right virtualization solution for your company, consider the following: • Don’t choose the “newest” or “cheapest” solution just because they are new and seemingly inexpensive. • Look for a solution that has been around for a long period of time to ensure the technology has been tested with a variety of applications • Look for a solution that has been proven in production IT environments • Choose a solution that offers flexibility and options to fit the needs of your company In my opinion, two solutions meet these criteria. They are VMware ESXi Free Edition and the VMware vSphere platform. The first is available for free and is a good way to start your company on the path toward virtualization. The second product can be evaluated for free and purchased as a low-cost package solution for smaller deployments. While these two solutions each have their own unique fit, theyboth have been proven by businesses of all sizes over a long period of time and they have the most to offer of any virtualization solution available today. The only way to truly get comfortable with virtualization is to try it for yourself on your own servers and perform tests in your environment. Download and evaluate any solution before making a purchasing decision.

Step #3 – Determine if applications are going to work well with virtualization.

One of the concerns I have heard from administrators who haven’t

virtualized their servers yet is that they believe their applications

might not be “virtualization friendly.” While there may be a few cases

in which this is true, the numbers of servers that can’t be virtualized

are small compared to the vast majority of all servers that can.

In my experience, if you understand the application, the majority

of the time, you won’t have any trouble consolidating a physical

server into a virtualized environment. I have successfully virtualized

Citrix Server, Exchange 2007 Servers, graphical applications,

database servers, and other critical enterprise applications. If you

are concerned that your virtual servers won’t offer the performance

that your applications demand visit the VMware Virtual Appliance

Marketplace (http://www.vmware. com/appliances/). Virtual

appliances are pre-built, preconfigured, ready-to-run enterprise

applications packaged with an operating system inside a virtual

machine.

Step #4 – Analyze the cost of virtualizing your server

infrastructure.

In Step #1, I mentioned that you can save your company

money by virtualizing your servers. As most of us work

for businesses, and businesses are in the business of

maximizing profits, it only makes sense that before

undertaking a virtualization project, you should analyze the

cost and potential savings (the ROI).

For those companies who value ROI, I anticipate that the

ROI of virtualization will always be there, but the question

may be how long does it take to achieve that ROI? When I

get asked that question, I recommend that you calculate

your ROI to virtualize your servers, with the VMware ROI

calculator. For example, when I used it, it showed that by

consolidating and virtualizing 20 physical servers down to

3, you could save $200,000 in server, related hardware and

power, cooling and real estate costs and $85,000 in IT staff

operating costs over 3 years1. Plus, if you have more physical

servers to start with, the cost savings are even greater.

Furthermore, the typical payback period, or amount of

time to break even on the investment, for a 20 server

consolidation project is 1 month. What other IT projects can

you say that about?

Even if you don’t use numbers, I believe that the cost

savings of virtualization is obvious. Virtualization requires:

• Fewer servers

• Fewer infrastructure costs – cooling, UPS, generator

• Less spent on electricity

• Less space needed for you IT infrastructure

• Less time spent administering servers

• Faster response to business needs

If you can install a product that does all those things, it will

eventually (and probably very quickly) pay for itself. That’s

what I call “a no-brainer.”

One point to note about comparing costs among

virtualization vendors. Some vendors like Microsoft and

Citrix will position their solutions as “free” compared with

VMware. We know that no solution that you rely on to stand

up your production infrastructure can really be “free.” Those

vendors have made their hypervisors free but shifted the

cost to their management tools, which are necessary to use

when managing a production environment. VMware has

introduced a method for comparing “cost per application,”

which they position as the true way to measure cost in an

apples-to-apples way. I would encourage you to explore

that cost comparison further.

Step #5 – Analyze the time and skill needed to virtualize your server infrastructure

I don’t want to minimize the time and skill required to

create a virtual environment. Depending on the scope of

the project, it could be very quick or it could be a more

significant undertaking. If I were to estimate the time to

learn about VMware vSphere and consolidate 20 physical

servers with “typical” applications onto two or three VMware

ESX Servers, it would look like this:

• Learn about VMware vSphere – via reading, video training,

or a VMware class – 1 week

• Install and configure VMware vSphere – 1 day

• Perform test server consolidations using VMware

Converter– 1 day

• Convert all 20 physical servers to virtual servers and

consolidate – 3 days

Total time = 2 weeks

Again, this is just a generalization with lots of assumptions

made. However, as you can see, being able to learn about

virtualization, get your virtual infrastructure installed

and configured, and consolidate 20 servers in a matter

of 2 weeks is a relatively small investment of time for

an effort that yields huge benefits and fast ROI for your

company. Keep in mind that VMware ESXi—the company’s

free solution that provides basic server optimization

functionality—requires even less time. You may even

choose to work with a VMware Partner who can assess

your requirements and install and configure VMware

virtualization software for you.

Make the move when you’re ready

Virtualization isn’t just for big companies anymore. There is

no doubt that you should virtualize your servers – it’s simply

a matter of when. I believe that the time to virtualize is now,

but you should make that call for yourself. Be sure to follow

the five steps outlined in this article before you make a

purchasing decision.

About the Author

David Davis is Director of Infrastructure at TrainSignal.com

– the global leader in video training for IT Pros. He has a

number of certifications including vExpert, VCP, CISSP, and

CCIE #9369. Additionally, David has authored hundreds

of articles and six different video training courses at Train

Signal with his most popular course being VMware vSphere.

His personal website is VMwareVideos.com. You can follow

David on Twitter or connect with David on LinkedIn.

VMware ESXi Design

The ESXi hypervisor shares many common elements with its older big brother ESX classic, but the main differentiator is that the Linux-based Service Console was stripped out. ESXi retains VMkernel and VMM components similar to ESX but has additional features built into the VMkernel; a new, much smaller management console; and other user-mode processes to replace the old Service Console OS functionality.

ESXi was redesigned this way to allow VMware users to scale out through a hypervisor
that is more akin to a hardware appliance. The vision was a base OS that is capable of autoconfiguring, receiving its settings remotely, and running from memory without disks. But it’s also an OS that’s flexible enough to be installed on hard disks along with a locally saved state and user-defined settings for smaller, ready-to-use installations that don’t require additional infrastructure.

Removing the Service Console obviously had an impact. A number of services and agents
that were normally installed had to be rethought. The familiar command-line interface with its
access to management, troubleshooting, and confi guration tools is replaced in ESXi. And the
Linux-styled third-party agents for backups, hardware monitoring, and the like must be provisioned in different ways.

VMware ESXi vs ESX (difference)

The removal of the Service Console fundamentally changed what was possible with VMware’s hypervisor. ESXi has a smaller and less demanding footprint, which means the hypervisor consumes fewer host resources to perform essentially the same functions. ESXi also uses significantly less disk space, whether local disk or boot-from-SAN space. The Service Console of ESX hosts effectively ran as a single vCPU guest, which meant all of its processes ran serially. With ESXi, those functions were moved to the VMkernel, meaning there is greater scope for those processes to run in parallel using much more sophisticated scheduling.

One unequivocal advantage of ESXi’s reduced code base is the greater level of security it brings. An ESX install came in at around 2 GB of installed fi les, whereas ESXi currently is near 125 MB. It’s easy to see that so much less code means less to keep secure with a smaller attack vector. The Service Console provided additional software that had to be secured, which ESXi avoids.


With fewer patches to apply, ESXi reduces the frequency of host-server reboots and lessens the administrative burden of regular patching. Any large enterprise with a sizable collection of ESX hosts will be only too familiar with the seemingly never-ending cycle of host patching. ESXi patches come as a single relatively small file, as opposed to ESX patches, which could be very large. Patching is also easier to manage if hosts are spread across several remote sites, particularly where slow WAN links cause issues with vCenter Update Manager’s (VUM’s) ability to push out these large packages. Another advantage with ESXi’s patches is that they come as a single firmware-like image that updates everything. Compare this to ESX patches, which came in multiple updates, potentially with dependencies on previous patches, and required multiple reboots.

An ESXi host is also more reliable than an ESX classic host. It effectively has less code to go wrong and fewer processes running over and above the VMs. The ability of the Service Console to run third-party agents was a double-edged sword, because although it allowed you to add extra functionality, some of the available agents caused stability issues on hosts. The inability of ESXi hosts to run unmanaged code means this is no longer a concern. Additionally, the dualimage nature of ESXi means there is always a standby bootable copy of the OS to roll back to, should you have any problems with an update.

ESXi brings with it the possibility of running hosts in a practically Stateless mode, meaning host servers are more comparable to hardware appliances. The deployment techniques available for ESXi are similar to those for ESX, but the install is considerably easier. You’re prompted for very little information, and the install time is incredibly short. You don’t need to understand the nuances of a POSIX fi lesystem and how best to carve up ESX’s partitions. Even rebooting an ESXi server takes considerably less time than rebooting an equivalent ESX classic host.

The simplifi cation of host management, with no need to understand a Service Console when confi guring and troubleshooting, means a lower entry bar for staff to install and maintain new vSphere environments. The simple Direct Console User Interface (DCUI) screen is more comparable to a BIOS setup screen and far less intimidating to staff unfamiliar with Linux. If a problem exists in a remote offi ce and there are no remote access cards or a keyboard, video and mouse (KVM) switch, then it’s more feasible that an onsite employee might be able to assist in restarting a management daemon.

ESXi has so many advantages that it’s clearly the better option for VMware and the company’s customers moving forward. Despite the many ESX classic installations that still exist awaiting migration to ESXi, clearly the smart option—the only option—is to deploy ESXi. Thanks to vSphere’s inherent abstraction, migrating VM workloads is relatively trivial and pain-free.

VMware vSphere Hypervisor

The term VMware vSphere hypervisor is actually a marketing term to specifi cally refer to the standalone version of ESXi that can be downloaded for free from the VMware website. It’s a restricted version of the hypervisor that can’t be managed by a vCenter instance. Remote connections via APIs are read-only, which limits third-party software such as backup and cloning tools, and there is a hard limit of 32 GB of physical memory for the host. It’s offered as a direct response to other free hypervisors on the market and is a useful springboard for many administrators just discovering virtualization.