Www.infotech.com Impact Research 1 The Business Case in Virtualization.

www.infotech.com Impact Research 1

The Business Case in Virtualization


Table of Contents

1. Executive Summary………………………………………………….. 3

2. Detailed Fit Indicators……………………………………………….. 6

3. Benefit Drivers………………………………………….………........ 13

4. Common Pitfalls…………………………………………………….. 16

5. Making the Virtualization Business Case……………………….... 21

6. Cost Avoidance……………………………………………………… 23

7. Cost Savings………………………………………………...…...…. 26

8. Necessary Expenditures……………………………………………. 29

9. Creating the Plan……………………………………………………. 30


Executive Summary


• Enterprise size affects the role of IT, the materiality of expenditure and savings, and comfort with new technologies.

• Very large companies may experience more difficulty in implementation due to established bureaucracies.

• Very small companies have trouble justifying the investment.

• Thirty servers or more leads to the greatest and most demonstrable cost savings and benefits when taking implementation costs into consideration

Mid-market enterprises are in the sweet spot to achieve cost savings from virtualization

Despite being a very powerful technology, virtualization benefits are more readily realized in certain types of enterprises.

Two key factors are leading indicators of greatest success:

If there is organizational fit, virtualization does provide most vendor-stated benefits.

Executive Summary

Size of Enterprise(# of employees)

Sweet Spot: ~100 to ~5,000 Sweet Spot: 15+

Number of Servers(# of Supported Physical Servers)


2. Reduction in hardware-related maintenance costs

The virtualization business case will be rooted in hardware savings

Implementation of Virtualization does not affect the number of applications, but rather the number of physical boxes in use.

Thus, the majority of pure cost savings are linked to reduction in physical boxes.

Cost avoidance and incremental business benefits exist in areas of Disaster Recovery (DR)/Business Continuity (BC) and efficiency, but are not always as tangible.

Executive Summary

1. Reduction in one-time and ongoing hardware acquisition costs

4. DR/BC enabled by removing need for

homogeneous hardware

3. Efficiency benefits from increased manageability

Business Case Value

40%-75%Acquisition Cost Savings

25%-50%Monthly Recurring Savings


Evaluating Your Company’s Fit


Detailed Fit Indicators

In the context of virtualization, success means achieving all three set benefits

SUCCESS

INTANGIBLE BENEFITS:

Recognized feature benefits inherent in virtualization and exhibited

improved efficiency.

TANGIBLE BENEFITS:

Achieved the stated business case financial objectives, including one-time and ongoing cost savings and on-time implementation

completion.

STRATEGIC BENEFITS:

Recognized value of virtualization and have plans in place to leverage

technology for future growth.


It’s necessary to look both within and outside of IT to accurately assess fit

Despite being a “cool” technology, perform adequate due diligence in assessing fit.


Virtualization can be implemented in any enterprise; however, success (as defined on the previous page) of that implementation depends on a number of factors, broken down into three categories:

CompanyCharacteristics

IT DepartmentCharacteristics

ServerInfrastructure


11%

67%

0%

10%

20%

30%

40%

50%

60%

70%

Small/Medium Enterprise Large Enterprise

Small and mid-sized companies make the most use of virtualization technology

Virtualization is really an infrastructure initiative that can bring some level of benefit to any size of company.

Although adoption penetration of virtualization is greater in larger companies, the proportion of their environment that is virtualized is much less significant.

The greatest benefit can be seen in small and mid-sized organizations.


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

1-100 101-500 501-1000

1001-2500

2501-5000

5001+

# Employees

Pe

net

ratio

n


Planned/POC

Using

Not Using

% o

f S

erv

ers

Vir

tualiz

ed


Smaller companies have seen greater benefits in the implementation of virtualization

Being a more innovative IT solution, small to mid-sized companies have significantly better success in implementation for the following reasons in order of impact (shown as Harvey Ball on right):


• Greater Executive Comfort with Virtualization:Comfort tends to be significantly higher in smaller organizations, in a large part due to the less complex political climate and ease of buy-in into the potential savings from a virtualization implementation. Further, smaller organizations tend to be less risk averse to new technologies than larger organizations.

• No Need for a Formal Approval Process:Flexibility of a smaller company, in most cases, eliminates the need for a formal approval process, which can introduce risk by necessitating education of a broader audience on the workings of virtualization.

• Reduced Political Complexity:Across organizations interviewed, the IT departments of smaller companies had more latitude to make decisions regarding virtualization and related infrastructure topics.

• IT Department Autonomy:Incremental savings resulting from benefits, such as maintenance efficiencies and power savings, have proportionally more impact on a small company (who feels the impact immediately) than a large enterprise.

Although never absent, fewer established (and isolated silos) and a flatter organization cause more pragmatic decision-making. The IT strategic vision can be politically diluted in larger companies.

Company Size(# Employees)

1-100

101-250

250-500

500-1000

1001-5000

5001+

Virtualization works, but is less compelling.

Sweet spot.

Virtualization works, but is more difficult to implement and usually of reduced scope.


Greater Executive Comfort with Virtualization

Reduced Political Complexity

Appetite for Incremental Cost Savings Relaxed Formal Approval Process

IT Department Autonomy

1

2

3

2

3


IT’s commitment to virtualization is key to reaping maximum benefits

IT departments’ current infrastructure and commitment to virtualization play a key role in determining success of an implementation.

Three characteristics are indicative of success:

IT DepartmentCharacteristics

• IT Comfort with Virtualization:Much like executive comfort with the technology, the technical team must be willing to move away from physical servers.

• Material Value of Power Savings:Facility space and power savings are two great benefits of implementing virtualization. If the IT organization views these as material, given their current environment and costs, then it indicates a good fit for the technology.

• Proportion of Servers to be Virtualized:Value in Virtualization is achieved when more, rather than less, of the server infrastructure is migrated. Migrating 50% or more of the environment is a good indicator of success.


IT Comfort with Virtualization

Material Value of Power Savings

High Proportion of Server Virtualization1

2

3


If you have 15 or more servers and plan to buy new virtual host machines, you’re in business!

The physical servers that exist in an enterprise are an important component of determining the fit of virtualization within the environment.

ServerInfrastructure

• 15 or more Physical Servers:The break-even point for virtualization of a small number of servers (considering hardware and software only) is roughly three servers. The value increases as the number of servers grow. Once all other costs are incorporated, 15 servers is the threshold to achieving greatest value. In short, the more, the better.

• Acquisition of New Physical Servers:Quick wins can be achieved by virtualizing low-utilization and low-complexity server applications like DNS and Web-servers. Critical applications are certainly candidates as well, but require more time to test thoroughly and may extend the payback timeframe.


• Number of Low-Utilization Applications:Reuse of physical servers to act as host machines is a great way to increase the utilization of those machines; however, for most implementations, purchasing new hardware will extend the life of host machines and could provide additional scalability.

15 or More Physical Servers

Number of Low-Utilization Application Servers

Please see Page 23for more details…

Acquisition of New Physical Servers1

2

3


Benefit Drivers


There are real savings resulting from virtualization implementation

Virtualization implementations bring benefits that span both tangible and intangible benefit areas:

• Cost Savings (spending less)• Cost Avoidance (not spending more)• Intangible Benefits (not quantifiable)

Many clients interviewed realized ongoing savings of

30-50%

SHRINK YOUR FOOTPRINTby decreasing rack space and cutting back on cooling/electrical costs

ENABLE AUTOMATIC FAILOVERto improve/enable DR and BC operations

HELP BUSINESS GROW FASTER WHILE KEEPING STAFF CONSTANTby making staff more efficient and repurposing/reallocating freed resources

DECREASE HARDWARE ACQUISITION COSTSby buying and setting up fewer servers

Co

st

Sa

vin

gs

Co

st

Av

oid

an

ce

Benefit Drivers






Intangible benefits can deliver as much value as hard cost savings

Virtual servers can be monitored and managed from one console and allow for capabilities such as failover and snapshot image backups. Features like these improve the efficiency of system administrators while maintaining, or improving, stability.

Products like VMware allow for resource pool management, which dynamically load balances virtual servers across physical hosts. The result is more consistent performance and less manual intervention.

Management of virtual servers should be accompanied by a cultural change that reduces or eliminates the purchasing of physical servers outside of the IT department. The IT department will provision virtual instances as required and hold the power to make decisions over incremental purchases.

Benefit Drivers

Improved Manageability & Flexibility

Improved Performance

Controlling Server Sprawl

Development projects are often slowed by the availability of development and testing environments. With template-like functionality, virtual servers can be brought up and refreshed from a single console within minutes to meet project requests.

Many companies are now virtualizing desktops to manage security (by allowing remote access to desktop images) and image integrity (through the ability to refresh a desktop image when necessary).

An interesting benefit to organizations running large, demanding applications on servers that require frequent hardware upgrades is that no re-installation or re-configuration is required upon hardware changes.

Perhaps not as salient, reducing cabling by five- or eight-fold does significantly reduce the possibility of human error in unplugging the wrong cable.

Shortened Timelines for Development Projects

Enabling Virtualization of Desktops

Separate Hardware from Software when Upgrading

Reduce the “Mess”

Inta

ng

ible

Be

ne

fits


Common Pitfalls


Heed the warnings of your peers (continued) Business case pitfalls

Be wary of including FTE headcount reduction

Companies we spoke to had found improved efficiency in their maintenance staff, but this is rarely sufficient to remove resources altogether since, in most cases, the system administrators perform other roles in the IT department.

Quell your expectation of huge consolidation. Consolidation ratios mentioned by the vendors differ from actual client experiences.

Common Pitfalls

Be Wary of Including FTE Headcount Reduction

Quell Your Expectation of HUGE Consolidation

0

3

6

9

12

Per ProcessorPer Core

What vendors report…

What to expect…

Nu

mb

er o

f V

irtu

al M

ach

ines


Heed the warnings of your peers (continued)

Common Pitfalls

Implementation warnings

Executives and Staff Need to have FaithDon’t underestimate the culture shift required to remove physical servers and replace them with virtual ones. “Server Huggers” (found in IT and in the business) need time to come on board and support the initiative.

You will most likely hit an IO BottleneckMemory and CPU can be managed and allocated to keep up with most virtual server demands. NICs can be scaled and added as necessary. I/O, however, will most likely be a bottleneck and the limiting factor as to the number of virtual machines a host can manage. For example, this is particularly true in Exchange environments with consolidated storage and a high number of BlackBerry users. From a capacity planning standpoint, one BlackBerry Exchange user translates into four standard Exchange users on an I/O basis.

Vendor Support may be an issueAlthough they are coming on quickly, not all vendors (especially smaller ones) are supporting their applications in a virtual environment. That being said, many enterprises we spoke to became reference clients for their software vendors once the virtual server implementation was complete. VMware currently has a number of initiatives underway to help gain software vendor support, one of which includes a group who works directly with ISVs. Of the top 70 ISVs, 75% now support VMware as a platform. Over time, vendor support will becomes less of an issue, as the vendors are recognizing virtualization’s prevalence in the market. Build in time to manage these issues with your vendor(s).

Executives and Staff Need to Have Faith

You Will Most Likely Hit an I/O Bottleneck

Vendor Support May Be an Issue Please see Page 36 for more details…



Common Pitfalls

Implementation warnings continued

VM Image Backups can be very largeAlthough a terrific feature, managing point-in-time snapshots of virtual machine images may not be worth the size requirement to make it beneficial.

Beware of Licensing CostsSome vendors may still be inclined to charge based upon the total available power of the host system, as opposed to the apportioned pool of resources allotted to each virtual machine. Investigate licensing in a virtual environment when assessing applications for migration.

Maintenance Contracts could be in jeopardyMuch like vendor support, hardware hosting outsourcers may not support environments running virtual servers. Build in time to renegotiate contracts or find new service providers.

Beware of Vendor Licensing Arrangements

Virtual Machine Image Backups Can Be Very Large

Maintenance Contracts Could Be in Jeopardy


Post-implementation considerations

Increased pay for staff due to skill setsOnce staff is trained in the implementation and maintenance of a virtual server environment, think about financial incentives since they will now possess a hot skill set. This could be akin to what happened with ERP skills in their prime, although not of the same magnitude. While this shouldn’t be a grave concern, include it in the business case as necessary.

Appropriate Security ConsiderationsServer management is made much easier through the implementation of a virtual server farm. However, this ease of management also requires appropriate controls and access permissions given the larger number of servers and scope of features that are being managed from one “pane”. Give few people access, and restrict features where feasible.


Common Pitfalls

The IT administrators at

one client location would

failover instances between

physical hosts many times a day because

“they could”.

Implement Appropriate Access Controls to Server Management Console

Increased Pay/Incentives for IT Staff’s New Skill Set


Making the Virtualization Business Case


The business case should be based primarily on hard cost savings

Virtualization has the benefit of having real cost savings; use these instead of quantifying other intangible benefits or cost avoidance items that may obfuscate the real cost-saving benefits which exist.

The preliminary business case can have a tolerance of +/- 25%, but be sure that all areas are addressed, as per the table below:

Business Case Areas

Ongoing

CostSavings

Costs

• Hardware Acquisition •Facilities: Cooling, Power and Rack Space

•Incremental Hardware Acquisition

•Disaster Recovery and Business Continuity

•Labor and Maintenance Savings

•Host Machine Hardware

•Virtualization Software

•Purchase of the SAN (if required)

•Staff Training and Consultancy Costs

•Virtualization Software Maintenance

•Staff Training (Turnover)

•Salary Increases (In-demand Skills)

One-time

CostAvoidance


Depending on the approach taken, host machines for virtualization will either be purchased new, or existing high-end servers can be reused:

• New server acquisition: the break-even point is collapsing three servers onto one virtual machine host assuming a two-way server configuration.

• Reusing a server: The server can be reused in an other part of the business OR a host purchase is not required, saving an additional $3,000-$6,000 and reducing break-even to a two-server consolidation.

• This does NOT include the purchase of a SAN if one is not available in your environment.

Including testing, labor costs, training and other implementation costs, the break-even point is at 15 physical servers or more.

Blade servers are complementary to virtualization as they provide a hardware-based redundancy.

The Total Cost of Ownership (TCO) of an eight-server blade environment is now 21% less than its equivalent rack-mount server installation.

Blades Are Viable Options As Well

Hardware acquisition savings are the main force driving the virtualization business case

Cost of New

Server$3-6K

0

$3000-$6000

$6000-$12000

$9000-$18000

Acquiring New Servers

Virtualizing

Cost of New

Server$3-6K

Cost of New

Server$3-6K

Cost of Virtual Host

Hardware$6-12K

Cost of Virtual

Software$3-6K

Hardware Acquisition Cost Savings


Break-even Is 3 Virtual Servers per Host Machine General RuleVirtual Server Break-Even

15 Servers Needed to Cover Implementation Costs


0%

25%

40%

50%

57%

63%

67%

70%

0% 10% 20% 30% 40% 50% 60% 70% 80%

3:1

4:1

5:1

6:1

7:1

8:1

9:1

10:1

Average % Hardware Cost Savings

Co

nso

lid

ati

on

Rat

io

The more servers that can be consolidated onto the hosts, the better the savings (one-time and ongoing)

z

Aprofessional

services client consolidated seven servers on each host

machine, realizing hardware savings of over 60%


If deemed material, facilities savings and cost avoidance are a key driving factor

Facilities savings come from two key areas:

• Power: server power and cooling Savings in power consumption come from a reduction in the number of physical servers and reduction in the air conditioning needs.

• Rack spaceWhether through simple virtualization or a combination of blade servers and virtualization, large amounts of rack space or raised-floor space are freed with server consolidation.

More than half of the clients interviewed state that shrinking the footprint was a driving force behind virtualization adoption.

However, this driver is less applicable to enterprises with unlimited space and large amounts of power in production, such as manufacturing companies.

Virtualization can help to cut power consumption in half, while freeing up considerable rack space in the server room.

Graycon, a mid-sized professional services firm, emptied 40% of its rack space, and plans to have 60%

emptied once implementation is

complete.

Facilities Cost Savings

Power Savings

Rack Space Availability

Bennett Jones, a large Canadian law firm, avoided the cost of

building a new facility through virtualization.


Enable high availability and DR

Heterogeneous Hardware:

No need for homogenous servers to achieve high availability/failover. Redundancy can be had without the need for “two of everything”.

Hardware Sharing:

Whereas applications, in the past, may not have coexisted within an environment, hardware can now be shared amongst competing applications as each will have its own dedicated environment. This enables failover to any available physical server.

Failover without user Impact:

Products like VMware’s VMotion allow for uninterrupted user experiences while moving instances between physical servers.

Backup/Restore: Snapshot Capability:

Point-in-time snapshots can be taken of an entire virtual machine image, minimizing the amount of time necessary to return an application to a particular state. Note here that applications that rely upon data on a SAN would only have the local storage contents restored.

Implementation of virtualization automatically enables some level of high availability and business continuity. Cost avoidance areas include:

Cost Avoidance in Disaster Recovery / Business Continuity

Coorstek, a large manufacturing company, was able to cut its

annual DR costs by over 35% because of

virtualization.

Heterogeneous Hardware

Hardware Sharing

Failover without User Impact

Backup and Restore – Snapshot Capability

Virtualization is software – hardware redundancy is still required.


Introduce efficient processes for server management

More Efficient Processes Eliminates delay related to new server delivery, set up, and takedown. Reduces development time by enabling the creation or refresh of development and

testing environments in minutes. Reduces time required to manage server farm by leveraging virtual machine

management tools that allow for single-pane control of multiple servers.

Efficiency leads to the ability to repurpose IT staff and get other initiatives underway which otherwise would have been postponed or incurred costs of contractor resources.

Many companies interviewed report a drop in new hires despite growth of the business and IT environment.

Reduce cost of hardware maintenance Fewer physical machines can mean a tangible reduction in hardware maintenance

contracts with server, or third-party, vendors

Better staff utilization through virtualization allows more efficient management of the IT department, repurposing of employees, and meeting of business growth demands.

Labour and Maintenance Savings

Clients stated that they are

able to keep IT staff constant, despite growth

in their business of up

to 50%.

More Efficient Processes

Repurpose Employees

Reduce Cost of Hardware Maintenance


Don’t look here for savings…

License costs The only benefit noticed was with clients who used the OS-hosted Microsoft solution,

which allows for four Microsoft servers to be run in virtual instances under one license. Otherwise the number of OS licenses remains constant.

Insurance costs reduction No noticeable benefits realized yet, but this may be something to watch for in the

future.

FTE reductions In most cases virtualization is not significant enough to justify head-count reduction

Low Benefit Areas

FTE Reductions

Insurance Cost Reduction

License Costs


Ensure the following key costs are included in the project budget

The cost side of the business case should contain the following items:

Hardware Acquisition of new servers to host virtual machines.

• Multi processor servers or blade servers. The purchase of a SAN.

Software Virtualization software server license.

Resources Training. Consultancy costs. Additional remuneration for virtualization-trained IT staff.

To achieve maximum benefit from the implementation, it is highly recommended that a SAN be included in the project budget.

Necessary Expenditures

Many companies interviewed stated that they were able

to use existing hardware as host

machines, or repurpose existing hardware in other

departments

Hardware

Software

Resources


Please see Page 42 for more details…



Creating the Plan


Twelve critical planning steps for virtualization implementation

Step 1: Determining the Approach

Step 2: Developing the Business Case

Step 3:Gaining Buy-In

Step 4: Capacity Planning &

Benchmarking

Step 5: Hardware Selection

Step 6: Software Selection

Step 8: Application Sequencing

Step 9:Testing

Step 10: Centralize/

ConsolidateStep 11: Migrate

Step 12: Monitoring & Expansion

Step 7: Resourcing

Based on successful cross-industry implementations, below is a compiled leading-practices guide to moving forward with virtualization.


Organization and project characteristics will determine the implementation approach

Step 1: Determining the Implementation Approach

IT Autonomy

HIGH LOW

High autonomy is characterized by: Executive trust in IT decision-making, resulting in a less formal

approval process Relatively high risk tolerance within the IT group Consolidated IT budget versus departmental IT budgets

Low autonomy is characterized by: Need for a formal business case and approval process Focus on hard cost savings and benefits as a result of C-level

understanding of IT Non-IT executive involvement and responsibility for IT projects

Scope of Implementation Scope of Implementation

LOW HIGH

Scope covers most systems

Project goals are to maximize cost savings

Minimizing overall upfront costs and risks

Mostly development and testing systems are tackled

Scope covers many critical systems

Many lines of business affected

IT-specific systems being affected (e.g. internal help desk, DNS servers, Web servers, etc.)

Be transparent in your plans. Gain executive approval and support.

Run as an internal IT project, and keep

business and executives informed.

Involve the business, and understand the full implications and Return

on Investment (ROI). Consider consultants.

Start small as an internal IT project. Learn from mistakes, and become more formalized with

time.

HIGH LOW


Type of implementation is influenced by company size

Small and mid-sized companies are more likely to virtualize a higher percentage of their servers and execute their projects as an IT-only Initiative.

Large Companies

Low

High

HighLow

Sco

pe

of

Imp

lem

en

tati

on

IT Autonomy

Mid-Sized Companies

Small Companies

Enterprise Companies

Step 1: Determining the Implementation Approach


Key components of the business case

1. Assess Fit: Assess whether the enterprise is an ideal candidate for virtualization based on fit indicators and benefit drivers described in this document and accompanying evaluation tool.

2. Determine Scope: Determine what servers and applications are candidates for migration.

3. Evaluate Alternatives: Make sure key categories, listed in the slides to follow, are covered: software and hardware selection (including the consideration of blades), the number of servers to consolidate initially, the implementation of a SAN, and training/external assistance.

4. Estimate Expenditures: Create a project budget by first identifying the necessary expenditures involved with the implementation.

5. Estimate One-Time and Ongoing Savings: Quantify the cost savings and consider the savings in cost avoidance that would be realized if they are tangible and recognized for the enterprise.

6. Estimate ROI: Determine the estimated ROI based on the above factors to be used for executive approval. Business cases typically have a +/- of 25, so make sure to caveat the estimates.

7. Develop Implementation Plan: Develop a timeline with key milestones for testing, training, implementation, and the date which the enterprise plans to revisit and expand the deployment.

While the depth of the business case will vary, it is important to consider all of the mentioned components prior to implementation.



General RuleDuration of Virtualization Implementation

Considerations that drive duration of implementation


Drivers Impacts

Number of Servers Being Migrated

As a General Rule, based on our analysis of various companies who have implemented from 15 to 200+ servers, a predictable correlation exists between the number of servers and duration.

Staffing and resourcing implications are discussed in Step 7.

Size of Organization

Generally, larger organizations have more formalized processes in place, which means acceptance testing and sign-offs will extend duration.

Criticality of Applications Being Migrated

The more critical the application, the more time should be set aside for testing. Consider regression, user acceptance, and performance testing scenarios.

11

1720

2326

14

0

5

10

15

20

25

30

15 30 45 60 75 90

# of Servers

Dur

atio

n (W

eeks

)

Please see Page 43 for more details…

Total Duration =2 Months + 1 Day per Migrated

Server


Considerations that drive duration of implementation (continued)

Almost all of interviewees stated that the project was met on time and was of anticipated complexity.


Drivers Impacts

Application Vendor Support

If the application vendor won’t support a virtual implementation, don’t think all is lost. In most cases we found the vendors came around with some proof-of-concept work and, in the end, used the account as a reference. Build in time for this “negotiation”.


Three key points to help gaining organization buy-in

APPOINT A TEAM LEADA formal team lead acts as the key contact for issues, questions, and communication distribution. Having a point-person helps to alleviate user concerns by matching a face to the project.

CONVINCE THE SERVER HUGGERSOvercome initial misconceptions about the technology through education. A proof-of-concept in a very effective way of getting quick on-the-job training for the IT staff and addressing some users concerns.

For executives, a well-positioned business case (with adequate contingency) is the best way to present the message as it is a compelling technology in that respect.

ROLL IT INTO AN APPLICATION RELEASEAs an alternative to a pure infrastructure project, a number of companies interviewed were successful in launching virtualization as part of a broader application release initiative (i.e. incremental servers).

Step 3: Gaining Organizational Buy-In

Depending on the approach that is most suited for the enterprise, deep buy-in may not be required, especially if the implementation is being run as in internal IT project.

A few successful clients have had great success in

putting together a formal presentation, in laymen’s terms, that describes the

technology and benefits.

Appoint a Team Lead

Bring the “Server Huggers” on Board

Roll It into an Application Release


Four-step process to determining capacity

Identify application servers that are candidates for migration:

Determine the average utilization rate, and make note of peak utilization rates. Ideally, a month’s worth of data would represent an adequate sample.

Choose the type of server that will host the virtual machines:

For most implementations, a two-way server is the minimum that should be considered given Hypervisor or OS overhead.

For higher ratio consolidation effort machines, four-ways deliver the best value. Refer to the chart on the next page for virtual machine-to-processor ratios of

successful implementations.

Step 4: Capacity Planning and Benchmarking

1

One-way Two-way Four-way Eight-way

Typically Underpowered

Typically Underpowered Price ProhibitivePrice Prohibitive

Best Value for High

Consolidation Ratios

Best Value for High

Consolidation Ratios

Ideal for Low-Utilization

Application Servers

Ideal for Low-Utilization

Application Servers

2

A large biotech firm hired consultants

for the explicit purpose of creating a capacity plan to

feed into the business case and

implementation plan.


1 2 3 4 5 6 7 8 9 10

Number of Virtual Machines per Processor

Low HighMedian

0

A good guideline is to consider three to four virtual machines per processor

Most companies have found a good mix of performance and consolidation value around three to four virtual machines per processor.

To be more accurate in planning, you can consider the number of virtual machines per processor core.



Four-step process for determining capacity (continued)

Set the target utilization rate for the host server considering failover capacity

Target utilization on host machines should be no more than 60% if the plan is to fail instances to it.

About 80% should be the maximum utilization on a host server in order to allow for adequate resource balancing for spikes in processing need.


3

4 Performance test to account for memory, NIC, and I/O constraints.

Be wary of any hard-stated recommendations of consolidation ratios or host server capacities.

Basin Power Electric Company went from having an average server utilization rate of

3% prior to virtualization, to

host machine utilization of 50%

with a 17:1 consolidation

ratio.


Blade servers are complementary to virtualization software and a viable choice

PROS • Can host large numbers of virtual machines in

four-way and eight-way configurations• Existing servers can be upgraded to act as host

machines

CONS• Can be costly to outfit with internal hardware

redundancy

Multi-processor Server Considerations

PROS• Have very small footprint and provide good

scalability (limited by chassis)• Provide some level of hardware redundancy

between blades

CONS• Run very hot for their size (dependent on number

of blades of course)

Blade Server Considerations

Step 5: Hardware Selection

While virtualization software provides redundancy at the OS level, blade servers can perform a similar function at a hardware level.

Consider the merits of both when making a decision.

Although not mandatory, a SAN will enable many of the benefits of a virtualization implementation. Make sure to include this expense in your calculations.


VMware is the current preferred product and market leader

Our interviews showed that the majority of companies are using VMware products, with a few running Microsoft Virtual Server.

Step 6: Software Selection

Vendors

Price Unsupported version is free Supported Enterprise

version 2-Socket license is $750

Free Hosted solution is free Hypervisor 2-Socket license

ranges from $1,000 to $5,750

Need for Host OS?

No – includes Hypervisor Yes – Windows Free version requires Windows or Linux

Licensed version includes Hypervisor

Supported Guest OSs

Linux, BSD in free version Windows, Linux, BSD, and

Solaris in Enterprise version

Windows, most Linux Windows, MS-DOS, Linux, Novell, FBSD, Sun Solaris

Benefits Open-Source product offers very reasonably prices tool-set

Install up to four copies of a server OS on one license

Mature tools for resource management and business continuity


Determine resource and staffing needs as this has potential impact on three types of costs:

• Training costs• Costs of backfilling seconded staff• Consultant costs

Estimate Staff Determine how many staff should be involved to

execute the implementation within the duration desired. A general rule for calculating staff requirements based on work effort appears on the right.

Ensure that project complexity and organizational risks are taken into consideration.

Determine Internal Availability and Capabilities Next, decide if the skills sets required are available

and the resources with those skills have capacity to take on additional tasks, or be seconded to the project full-time.

Start by estimating the total number of resources required

Step 7: Resourcing

General RuleWork Effort Required for Implementation

Total Work Effort (in Person Weeks) = (2.9 x # of Servers) -31

Staff Requirements =Total Work Effort (in Person Weeks)

# of Weeks of Desired Duration

Number of Servers

Duration in Weeks

Staff (FTE)

15 4 3.1

15 8 1.6

30 8 7.0

30 16 3.5

60 16 8.9

60 26 5.5

Estimate Staffing Needs

Determine Availability and Capabilities of Staff


Once staffing needs, resource availability, and capabilities have been identified, consider the mix of training and external consultants that you will leverage.

Our interviews led us to the conclusion that staffing mix and training are strongly correlated to the size of company, not necessarily the size of the implementation:

Small CompaniesGiven a smaller staff with varied duties, small companies tended to allow their staff to learn on the job while executing the project.

Medium-sized CompaniesThese companies would send one or two resources to training to become the trainers for the rest of the staff.

Large CompaniesIn most cases, given the transparent nature of the project, and the various parties involved, consultants were brought in to provide guidance and ensure an expedient implementation.

Company size is a good indicator of what type of training or external assistance you should investigate

On-the-job

training

Low

High

Small Medium Large

External

training

On-the-job

training

On-the-job

training

Consul-tants

External

training

Size of Company

Cost

of

Reso

urc

ing

Step 7: Resourcing

Small Companies

Medium-Sized Companies

Large Companies


Key considerations when evaluating training and external consultant alternatives

The larger your implementation, the proportionally less relevant the training and consultancy costs.

Staff Training Costing Questionnaire

If not, consider consultants and/or new hires.

1Do my staff have the understanding to build these new skills?

3 4

Consulting Costing Questionnaire

3

1 22

Do I need to backfillstaff on training?

Do I have high turnover?Should I include a salary premium?

Do I need consultants for point expertise to complement your team?

Do I need consultants to run the project?

Do I need the consultants to augment my team under the assumption I can’t meet mydeadlines with current staff?

If so, what is the cost?

If so, consider the recurrence of training costs.

If training, consider that these staff are now more valuable in the market

10-20% of total project time is probably adequate.

No more than 80%. Ensure a projectmanager comes with the consultant staff.

Maximum 80% staffingof consultants is recommended. internal staff should be involvedto some extent. Recommended to keep the mix around 50%.

Step 7: Resourcing


Investigate vendor licensing in a virtual environment first – some vendors may still be inclined to charge based upon the total available power of the host system, as opposed to the apportioned

pool of resources allotted to each virtual machine.

Step 8: Application Sequencing

Migrate critical applications first when:

A business case has been created and formal

approval process completed. In this scenario, business representatives are involved at the appropriate levels to gain buy-in for a critical application.

Improving DR capabilities is top concern.

Consultants have been brought on board to guide or execute the implementation. Their experience should be leveraged towards critical application migration.

CRITICAL APPLICATIONS LOW-UTILIZATION INTERNAL IT APPLICATIONS

Note: Not every service can be virtualized and some services require special attention in order for virtualization to work.

Migrate low-utilization internal IT apps first when:

The project is being run as an under-the-radar internal IT project with goals to minimize risk.

Cost savings is the main concern.

Gaining buy-in and building faith in the technology are key goals, with plans to move to more critical systems later.

Virtualize low-utilization applications if the project is an internal IT initiative


Given the nature of this technology, adequate stability and performance testing is a must

Info-Tech recommends purchasing a SAN when implementing virtualization.

Step 9: TestingStep 10: Centralize and Consolidate

Two types of testing are critical to the success of a virtualization implementation:

Regression/ Stability TestingThe testing of an application’s behavior in a virtual environment cannot be predicted and must be tested thoroughly before production launch

Performance TestingEven with a detailed capacity plan, some performance testing and tuning must occur in the testing environment. Resource allocation and load balancing features must be configured and tested to handle peaks.

Virtualization is a consolidation exercise and, as such, centralization of servers is key to achieving favorable consolidation ratios.

The majority of our interviewed companies were already centralized, or centralized a portion of their infrastructure prior to virtualization.

We did find, however, that mini-business cases can be created at a branch level given a critical mass of servers at each location, and a justification (either bandwidth or staff location) for not moving servers away.

Regression/Stability Testing

Performance Testing / Tuning

Centralize First

Unless you have a Business Case Not to


Expanded Services

Partnerships

Acquisitions

Status Quo

Exit / Sell

Organizational Buy-in

Business Case Development

Implementation Approach

Capacity Planning

Hardware/Software Selection

Pre-Implementation ResultsImplementation Steps

AverageImplemen-

tationTakesSeven Months

StaffTraining Testing

Centralize and

Consolidate

Migrate

Time to Implement

5%

15%

40%40%

Step 11: Migrate

Percentages listed below are rough guides for time allocation across the steps. Pre-implementation activities will take longer in larger organizations than smaller ones. The key to success is getting into the implementation steps quickly and working out the issues in a testing cycle. Once testing is complete, the migration may take the same or less time to complete.

Roughly 85% of your time should be spent on implementation steps


Virtualization implementation is an ongoing process

Once the initial implementation is complete, key processes and plans need to be put into place:

Roadmap of Applications to MigrateTo continue the benefits stream, a roadmap of future application servers to migrate should be created (assuming the initial implementation migrated a subset of servers).

Process for Server RequestsVirtual server sprawl is as much a problem as physical server sprawl. A request process should be put into place to manage server instances.

On-Going Capacity Planning As additional instances are migrated and as application dynamics change, constant monitoring of resource pools is required to ensure performance targets are being met. This is especially important as physical hosts are targets for DR of failover instances.

Cross-TrainingSystem administrators will now have control over a larger pool of server instances. To ensure succession planning in the event of turnover or absence, a knowledge transfer and cross-training plan should be put into place post-implementation.

Collect Business Case MetricsTo maintain faith in IT’s abilities to deliver value, close the loop and collect post-implementation metrics to validate the original business case.

Step 12: Monitoring and Expansion

Create a Roadmap of Application Servers to Migrate

Establish Process for Server Instance Requests

Begin Cross-training Immediately

Continue Ongoing Capacity Monitoring and Planning

Collect Original Business Case Metrics


Implementing virtualization can be an entry into utility computing

Despite virtualization being an infrastructure initiative, there are process and strategy ramifications on how IT can/should operate with the new flexibility of virtualization.

Chargebacks and Transition to Utility ComputingVirtualization inherently lends itself to utility computing, where each virtual instance, the memory, CPU, and disk it uses are procured as required and charged for. Chargeback models, if a cultural fit for your enterprise, are complementary to virtualization in that measurement of chargeback metrics is made easier as is the provisioning of capacity to meet various Service Level Agreements (SLAs).

Step 12: Monitoring and Expansion

Transition to Utility Computing and Chargebacks

Www.infotech.com Impact Research 1 The Business Case in Virtualization.

Documents

virtualization benefits

virtualization slide

virtualization implementation

impact research

virtualization business

context of virtualization

workings of virtualization

greater benefits