This presentation, including any supporting materials, is owned by Gartner, Inc.

and/or its affiliates and is for the sole use of the intended Gartner audience or other
authorized recipients. This presentation may contain information that is confidential, proprietary or otherwise legally protected, and it may not be further copied,
distributed or publicly displayed without the express written permission of Gartner, Inc. or its affiliates.
2012 Gartner, Inc. and/or its affiliates. All rights reserved.
Nik Simpson
Comparing Data Center Costs With
Public IaaS Cloud Services
Summary of Findings
Comparing Data Center Costs With Public
IaaS Cloud Services
Bottom Line
- Comparing internal application hosting costs with costs for running
the application in an infrastructure as a service (IaaS) cloud
provider isn't easy.
- Many organizations have limited understanding of real IT costs,
and gaining that understanding requires information about all
aspects of the infrastructure and operational practices that is often
not measurable without significant effort.
- Because of the difficulty of creating an accurate cost model,
organizations should consider carefully whether the effort is
justified.
- Regardless of whether a direct comparison is made, organizations
should always investigate the costs required to prepare for
supporting applications in the external cloud, which are factors that
many organizations overlook.
1
Comparing Data Center Costs With Public
IaaS Cloud Services
Context
- IaaS cloud providers are an alternative to conventional IT
infrastructure for some enterprise workloads. As a result, many IT
organizations are considering moving some workloads out of their
data centers or using the cloud for new applications and so are
naturally interested in how the costs will compare.
- Unfortunately, many of these organizations have difficulty
calculating internal costs with any degree of accuracy and are
often misled by cloud cost calculators that oversimplify the
problem.
- In the past, IT organizations have been faced with similar problems
related to large-scale outsourcing of IT functions, where expected
cost savings proved to be an illusion, and are determined not to
make the same mistakes again.
2
Comparing Data Center Costs With Public
IaaS Cloud Services
Take-Aways: IT organizations should consider several
important take-aways when trying to compare internal
costs and cloud-hosting costs:
- Calculating internal costs for a particular application with any
degree of accuracy is a complex and time-consuming task. The
information required for even the most basic cost modeling will
come from multiple sources within the organization and will be
difficult to assign correctly, especially for shared infrastructure and
services. If a cost model is to be built, it will require cooperation
across the organization, including HR, facilities, purchasing and
the IT organization. Therefore, high-level sponsorship of the effort
is required. As a result, organizations that don't already have an
accurate cost model should decide whether the effort required to
create such a model is worth the energy and effort, particularly if
factors other than cost are more important in the final decision.
3
Comparing Data Center Costs With Public
IaaS Cloud Services
Take-Aways: IT organizations should consider several
important take-aways when trying to compare internal
costs and cloud-hosting costs:
- Calculating internal costs is an inexact science at best and may
produce an inaccurate model that yields estimates for internal cost
with significant margin of error. The real danger of an inaccurate
model is that incorrect decisions may be made, and those
decisions will be given a false legitimacy, thus making them hard to
undo in the future. Therefore, cost modeling should only be
attempted if there is a clear understanding of the complexity and
limitations of a data center cost model.
- Many organizations find that unexpected ancillary costs related to
moving applications to the cloud (for example, additional network
costs, costs to reintegrate cloud-based and local applications, staff
training, management integration, and identity services), rather
than running costs in the cloud, lead to cost problems. Therefore,
IT organizations need to understand these costs in detail as part of
the planning and budgeting process.
4
Comparing Data Center Costs With Public
IaaS Cloud Services
Take-Aways: IT organizations should consider several
important take-aways when trying to compare internal costs
and cloud-hosting costs:
- Cloud IaaS service providers have many cost advantages over a typical IT
organization, including scale; homogeneity; modern, purpose-built energy-
efficient facilities; and automation. As a result, their internal cost to operate
their infrastructure will almost certainly be lower than that of a typical IT
organization. However, the cloud provider still has to make a profit, so the
cost of service incurred by the IT organization will not fully reflect a cloud
provider's internal cost advantages. In addition, the day-to-day costs of
running the application in the external cloud (that is, the cost of the virtual
machine [VM] and any other resource-consumption-related charges for
networks and storage) are only a part of the overall costs. Organizations
must also factor in a variety of other costs, including investments in new
management software, changes to internal and external networks,
training, and a host of other ancillary costs. Therefore, organizations
should be wary of placing too much emphasis on cost savings as a reason
to move to the cloud.
5
Comparing Data Center Costs With Public
IaaS Cloud Services
Take-Aways: IT organizations should consider several
important take-aways when trying to compare internal
costs and cloud-hosting costs:
- Most public cloud providers offer online calculators for estimating
the pay-as-you-go (PAYG) running costs for an application hosted
in the cloud. However, care should be taken with these calculators
because they only cover day-to-day running costs and often
require the user to make guesses about such things as network
bandwidth consumption. In addition, these calculators don't
address costs incurred in preparing the infrastructure, people and
processes to support applications in the cloud. In addition, these
calculators fail to reflect ongoing costs for such things as
management, integration, support, network transmission, disaster
recovery (DR), performance monitoring, personnel, training,
licensing and dependencies. As a result, these calculators often
seriously underestimate the cost of moving an application to, and
then managing it in, the cloud.
6
Comparing Data Center Costs With Public
IaaS Cloud Services
Take-Aways: IT organizations should consider several
important take-aways when trying to compare internal
costs and cloud-hosting costs:
- PAYG pricing has become synonymous with public cloud.
However, PAYG isn't the best option in many circumstances. As a
result, many IaaS cloud providers now offer a variety of options to
customers, including fixed price contracts and other payment
systems, to eliminate the unpredictable fluctuations of a traditional
PAYG approach.
- For PAYG service, calculating the long-term running costs in the
cloud will require detailed understanding of how the cloud provider
charges for service and how those charges map to the application
being relocated. For example, charges for network traffic from the
application, or data storage charges, may be a major and variable
component of the final bill. Estimating these costs beforehand may
require instrumenting of the application to understand such things
as network traffic flows while the application still runs in the data
center.
7
Comparing Data Center Costs With Public
IaaS Cloud Services
Take-Aways: IT organizations should consider several
important take-aways when trying to compare internal costs
and cloud-hosting costs:
- IT organizations making their initial foray into public cloud IaaS should
always start with a pilot project for a single application and gather cloud
cost data for a few months before considering broader deployments.
Starting small has the benefit of allowing a company to get more
experience with the true costs of running applications in the cloud without
making a major commitment.
- There are several issues that IT takes of care of in-house (for example,
supporting a legacy application that requires an old release of an
operating system or the use of an application version that a vendor no
longer supports). Organizations cannot assume that external cloud
providers will be accommodating if they decide to release a significant
upgrade of their hosting platform or make radical changes to a service that
breaks an application. Cloud providers will strive for uniformity across their
infrastructure, and making exceptions for individual customers is not part
of their operating model.
8
Comparing Data Center Costs With Public
IaaS Cloud Services
Conclusion
- Building an internal cost model isn't easy for many
organizations. However, assuming that external cloud will be
cheaper is a dangerous assumption to make; moving to the
cloud may well prove more expensive than continuing with
the existing approach. Therefore, it's important to
understand just how meaningful internal versus external
costs are to the overall decision before embarking on a cost-
modeling exercise.
- Regardless of whether an internal versus external
comparison is required, organizations should always
understand the long-term costs of operating an application in
the cloud and the costs incurred by moving an application to
the cloud.
9