Cloud computing and sustainability: The environmental benefits of moving to the cloud - a look at the future of the cloud and it's 'greener' motives.

1. Cloud Computing and Sustainability:
The Environmental Benefits of Moving
to the Cloud
In collaboration with

2. Executive Summary 2 Case Study – Global Consumer
Introduction: Is the Cloud a Goods Company 10
“Greener” Computing Alternative? 3 Conclusion & Outlook 11
Research Approach 4 Expanding the Cloud 11
Summary Findings 5 Further Improvements 11
How Does Cloud Computing Reduce Appendix 12
Environmental Impacts of IT? 6
Dynamic Provisioning 6
Multi-Tenancy 7
Server Utilization 7
Data Center Efficiency 8
Other Important Factors 8
2

3. Executive Cloud computing—large-scale, shared
IT infrastructure available over the
Several key factors enable cloud
computing to lower energy use
Summary internet—is transforming the way
corporate IT services are delivered and
and carbon emissions from IT:
• Dynamic Provisioning: Reducing
managed.
wasted computing resources
To assess the environmental impact of through better matching of server
The cloud’s cloud computing, Microsoft engaged capacity with actual demand.
with Accenture—a leading technology,
unprecedented economies consulting and outsourcing company—
• Multi-Tenancy: Flattening relative
peak loads by serving large numbers
of scale reduce overall and WSP Environment & Energy—a
of organizations and users on
global consultancy dedicated to
cost and increase environmental and sustainability
shared infrastructure.
efficiencies, especially issues—to compare the energy use • Server Utilization: Operating
and carbon footprint of Microsoft servers at higher utilization rates.
when replacing an cloud offerings for businesses with • Data Center Efficiency: Utilizing
organization’s locally corresponding Microsoft on-premise advanced data center infrastructure
deployments.
operated on-premise1 designs that reduce power loss
The analysis focused on three of through improved cooling, power
servers. But does this Microsoft’s mainstream business conditioning, etc.
advantage also translate applications—Microsoft Exchange®, Though large organizations can
Microsoft SharePoint® and Microsoft
to environmental Dynamics® CRM. Each application
lower energy use and emissions
by addressing some of these factors
benefits? is available both as an on-premise in their own data centers, providers
version and as a cloud-based of public cloud infrastructure are
equivalent.2 The team compared best positioned to reduce the
the environmental impact of cloud- environmental impact of IT because of
based vs. on-premise IT delivery on their scale. By moving applications to
a per-user basis and considered three cloud services offered by Microsoft
different deployment sizes—small or other providers, IT decision-
(100 users), medium (1,000 users) makers can take advantage
and large (10,000 users). of highly efficient cloud
The study found that, for large infrastructure, effectively
deployments, Microsoft’s cloud “outsourcing” their IT efficiency
solutions can reduce energy use investments while helping their
and carbon emissions by more than company achieve its sustainability
30 percent when compared to their goals. Beyond the commonly cited
corresponding Microsoft business benefits of cloud computing—
applications installed on-premise. such as cost savings and increased
The benefits are even more impressive agility—cloud computing has the
for small deployments: Energy use potential to significantly reduce
and emissions can be reduced by the carbon footprint of many
more than 90 percent with a shared business applications.
cloud service.
2

4. Introduction: Is the Cloud a “Greener”
Computing Alternative?
Both cloud computing and Economy In the Information Age” Note: While this research focuses
sustainability are emerging as estimates that the environmental on direct carbon reduction benefits
transformative trends in business footprint from data centers will more of the cloud, it is also important to
and society. Most consumers than triple between 2002 and 2020, mention potential indirect benefits
(whether they are aware of it making them the fastest-growing of cloud computing beyond the
or not) are already heavy users of contributor to the Information and scope of this study. Like broadband
cloud-enabled services, including Communication Technology (ICT) and other technologies provided
email, social media, online gaming, sector’s carbon footprint. by the ICT sector, cloud computing
and many mobile applications. is emerging as a viable, scalable
It stands to reason that consolidating
The business community has begun technology that can help significantly
corporate IT environments into large-
to embrace cloud computing as a reduce carbon emissions by enabling
scale shared infrastructure operated
viable option to reduce costs and new solutions for smart grids, smart
by specialized cloud providers would
to improve IT and business agility. buildings, optimized logistics and
reduce the overall environmental
dematerialization. The Smart 2020
At the same time, sustainability impact and unlock new efficiencies.
report estimates the potential impact
continues to gain importance But does this assumption pass the
of ICT-enabled solutions to be as
as a performance indicator test of a quantitative assessment
much as 15 percent of total global
for organizations and their on a per-user basis?
carbon emissions (or 7.8 billion tons
IT departments. Corporate Considerable research has been of CO2 equivalents per year). Broad
sustainability officers, regulators dedicated to understanding the adoption of cloud computing can
and other stakeholders have environmental impact of data stimulate innovation and accelerate
become increasingly focused on IT’s centers and to improving their the deployment of these enabled
carbon footprint, and organizations efficiency.5 However, the aggregate solutions. Consequently, cloud
are likewise placing more emphasis sustainability impact of choosing computing may have a major impact
on developing long-term strategies a cloud-based application over an on global carbon emissions through
to reduce their carbon footprint on-premise deployment for the same indirect benefits in addition to the
through more sustainable operations application has not been rigorously direct savings from replacement
and products.3 analyzed. For example, how might a of on-premise infrastructure
Cloud service providers are making CRM solution for 1,000 sales agents which are analyzed here.
significant investments in data reduce the overall environmental
center infrastructure to provide not footprint when it is run in the cloud
only raw computing power but also versus on a company’s own servers?
Software-as-a-Service (SaaS) business Is there a net benefit of moving
applications for their customers. to the cloud, or are we simply
New data centers are being built at “outsourcing” the environmental
ever-larger scales and with increased impact to a service provider? This
server density, resulting in greater Microsoft-sponsored study is targeted
energy consumption. The Smart 2020 at answering these kinds of questions.
report4 “Enabling the Low Carbon
3

5. Research Approach
Building upon previous analysis work • Power Consumption for To assess the carbon footprint
with Microsoft,6 Accenture and WSP Networking8 and Storage: Average of cloud-based applications, the
Environment & Energy developed a power consumed for networking research team collected data from
quantitative model to calculate the and storage equipment in addition Microsoft’s current data center
energy use and carbon footprint of an to server power consumption. operations. On-premise deployments
organization’s IT applications for both were modeled based on Microsoft’s
• Data Center Power Usage
cloud and on-premise deployment. product recommendations and input
Effectiveness (PUE): Data center
This approach aligns with the from subject-matter experts, and
efficiency metric which is defined
assessment methodology developed by were validated with a case study
as the ratio of the total data
the Global e-Sustainability Initiative using actual customer data.
center power consumption divided
(GeSI),7 the industry consortium
by the power consumption of The assessment looked at the
promoting sustainability on behalf of
the IT equipment. Power usage environmental impact of three
leading ICT companies.
effectiveness accounts for the different Microsoft applications—
The model quantifies energy use power overhead from cooling, all of them major products in
and carbon emissions on a per-user power conditioning, lighting and Microsoft’s portfolio of (server-
basis. To account for the fact that other components of the data based) business applications:9
on-premise server counts do not center infrastructure.
• Microsoft Exchange (email,
follow a linear scale as user counts
• Data Center Carbon Intensity: calendar and contacts)
increase, the research analyzes the
Amount of carbon emitted to
impact among three different sizes • Microsoft SharePoint (collaboration
generate the energy consumed
of user groups: small (100 users), and web publishing)
by a data center, depending
medium (1,000 users) and large • Microsoft Dynamics CRM (customer
on the mix of primary energy
(10,000 users). relationship management)
sources (coal, hydro, nuclear,
Specific input data utilized by the wind, etc.) and transmission These products are representative of
research team included the following losses. The carbon emission factor three types of business applications
(also see the Appendix for more is a measurement of the carbon that are used broadly by companies
detailed information): intensity of these energy sources. across industries. Assessing multiple
• User Count: Number of provisioned applications with different usage
users for a given application. characteristics provides a diverse
set of data points to validate the
• Server Count: Number of
hypothesis.
production servers to operate
a given application.
• Device Utilization: Computational
load that a device (server, network
device or storage array) is handling
relative to the specified peak load.
• Power Consumption per Server:
Average power consumed by a
server.
4

6. Summary Findings
The results of the analysis for As the data shows, the per-user Note that, because Microsoft
Microsoft clearly show significant energy use and carbon footprint is applications and data centers were
decreases in CO2 emissions per heavily dependent on the size of the the basis of the study, the specific
user across the board for cloud- deployment. The cloud advantage carbon reductions from running other
based versus on-premise delivery is particularly compelling for small applications from other software
of the three applications studied deployments, because a dedicated providers on a cloud model may
(see Figure 1). infrastructure for small user counts— vary. However, the trends shown
as in a small business running its own here are instructive and may be
The analysis suggests that, on average
servers—typically operates at a very used as directional indicators for
across the different applications,
low utilization level and may be idle decision makers in corporate IT and
typical carbon emission reductions
for a large part of the day. However, sustainability leadership positions
by deployment size are:
even large companies serving when considering a switch to cloud
• More than 90 percent for small thousands of users can derive computing with any provider.
deployments of about 100 users efficiencies from the cloud beyond
• 60 to 90 percent for medium-sized those typically found in on-premise
deployments of about 1,000 users IT operations.
• 30 to 60 percent for large
deployments of about 10,000 users
Figure 1: Comparison of Carbon Emissions of Cloud-Based vs. On-Premise Delivery of Three Microsoft
Applications
Microsoft Exchange Microsoft Sharepoint Microsoft Dynamics CRM
On-premise vs. Cloud Comparison, On-premise vs. Cloud Comparison, On-premise vs. Cloud Comparison,
CO2e per user CO2e per user CO2e per user
>90% >90% >90%
>90%
79%
81% 76%
52% 20%
Small Medium Large Microsoft Small Medium Large Microsoft Small Medium Large Microsoft
On-Premise On-Premise On-Premise Cloud On-Premise On-Premise On-Premise Cloud On-Premise On-Premise On-Premise Cloud
= estimated decrease with Microsoft Cloud
5

7. How Does Cloud Computing Reduce
the Environmental Impact of IT?
To understand the potential a reduced need for IT infrastructure kind of over-provisioning typically
advantage of cloud computing in to support a given user base. In turn, results from:
more detail, it is important to look at these primary levers are heavily
• The desire to avoid ongoing capacity
the distinct factors contributing to a influenced by four key factors (see
adjustments as demand fluctuates
lower per-user carbon footprint. These Figure 2):
over time.
factors apply across cloud providers
• Dynamic Provisioning
in general and are even relevant for • Difficulty in understanding and
many on-premise scenarios. This level • Multi-Tenancy predicting demand growth and
of understanding can thus help IT peak loads.
• Server Utilization
executives target additional efficiency • Budget policies that encourage
gains in an on-premise environment • Data Center Efficiency (expressed by
using all available funds in a given
and realize additional performance power usage effectiveness)
year to avoid smaller allocations
advantages in the future. Dynamic Provisioning the following fiscal year.
Generally speaking, the comparatively IT managers typically deploy far
Over-provisioning is certainly
smaller carbon footprint of cloud more server, networking and storage
understandable. Application
computing is a consequence of both infrastructure than is actually needed
availability is a high priority in IT
improved infrastructure efficiency and to meet application demand. This
operations, because IT executives
want to avoid situations in which
Figure 2: Key Drivers of Cloud Computing’s Reduced Environmental Footprint
Cloud Benefits
Reduce over-allocating Forecasting and ongoing adjustment of allocated
of infrastructure capacity avoids unnecessary over-allocation of
(Dynamic Provisioning) resources and sizing close to actual usage.
Share application Sharing application instances between client
instances between organizations (tenants) flattens peak loads and
multiple organizations reduces overhead for tenant onboarding and
Reduce Carbon
(Multi-Tenancy) management.
Footprint per User
Operate server Large deployments of virtualized server infrastructure
infrastructure at serving multiple tenants can balance compute and
higher utilization storage loads across physical servers and thus be
operated at higher utilization rates.
Improve data center Industrialized data center design at scale and
efficiency (PUE) optimized for power efficiency reduces power wasted
on cooling, UPS etc. and allows running servers at
optimal utilization and temperature.
6

8. business demand for services multi-tenant environments, cloud far fewer servers. Whereas a typical
exceeds what IT can provide. providers can reduce overhead for on-premise application may run at
Thus infrastructure planning on-boarding and managing individual 5 to 10 percent average utilization
is typically conducted with a organizations and users. rate, the same application in the
conservative, “just in case” mindset cloud may attain 40 to 70 percent
The Microsoft cloud offerings
that results in capacity allocation that utilization, thus dramatically
analyzed in this study are relatively
is not aligned with actual demand. increasing the number of users
new and are currently experiencing
served per machine.10
By contrast, cloud providers tend rapid growth. The more mature a
to manage capacity much more given cloud service becomes, the It is important to note that while
diligently, because over-provisioning less the demand will fluctuate, servers running at higher utilization
at the cloud’s operational scale resulting in even greater energy rates consume more power, the
can be very expensive. Providers savings in the future. resulting increase is more than offset
typically have dedicated resources by the relative performance gains. As
to monitor and predict demand and
Server Utilization illustrated in Figure 3, increasing the
Cloud computing can drive energy
continually adjust capacity, and utilization rate from 5 to 20 percent
savings by improving server utilization
their teams have developed greater will allow a server to process four
(the measurement of the portion
expertise in demand modeling and times the previous load, while power
of a server’s capacity that an
in the use of sophisticated tools consumed by the server may only
application actively uses). As large-
to manage the number of running increase by 10 or 20 percent.11
scale cloud providers tend to run
servers. Thus, cloud providers can
their infrastructure at higher and Virtualization offers a strategy to
reduce the inefficiency caused by
more stable utilization levels than improve server utilization for both
over-provisioning by optimizing the
corresponding on-premise operations, cloud and on-premise scenarios
number of active servers to support
the same tasks can be performed with by allowing applications to run in
a given user base.
Multi-Tenancy
Just as multiple tenants in an
apartment building use less power
overall than the same number of Figure 3: Relationship between Server Utilization
people owning their own homes, and Power Consumption
so do the multiple tenants of a
cloud-provided infrastructure
reduce their overall energy use 100%
and associated carbon emissions.
The cloud architecture allows Minor increase
providers to simultaneously serve in power consumption
multiple companies on the same Power
server infrastructure. Disparate Consumed Factor 4
demand patterns from numerous performance
increase
companies flatten overall demand
peaks and make fluctuations more
predictable. The ratio between peak
and average loads becomes smaller,
0% 5% 20% 100%
and that in turn reduces the need
for extra infrastructure. Major cloud Computational Load (Utilization)
providers are able to serve millions
of users at thousands of companies
simultaneously on one massive
shared infrastructure. By operating
7

9. an environment separated from has released an update13 to its initial Other Important Factors
the underlying physical servers. 2007 Report to Congress,14 stating an In addition to the four primary drivers
Multiple virtual machines can share average power usage effectiveness of of cloud computing’s environmental
a physical server running at high 1.91 for U.S. data centers, with most advantage, other contributing factors
utilization, which reduces the number businesses averaging 1.97. are worth mentioning:
of physical servers required to meet
Through innovation and economies • Hardware comes with an
the same demand. IT organizations
of scale, cloud providers can “embodied” carbon footprint
can scale individual virtual resources
significantly improve power usage from the energy associated with
to fit application needs instead of
effectiveness. Today’s state-of-the- producing, distributing and
allocating an entire physical system
art data center designs for large disposing of equipment. For the
whose full capability is not utilized.
cloud service providers achieve scenarios analyzed, this energy
In this way, virtualization provides a
power usage effectiveness levels outlay adds about 10 percent to
tool for IT departments to narrow the
as low as 1.1 to 1.2.15 This efficiency the footprint from IT operations.
efficiency gap between on-premise
gain could reduce power consumption The total hardware impact depends
deployment and a multi-tenant cloud
over traditional enterprise data heavily on the type of equipment,
service.
centers by 40 percent through data refresh cycles and end-of-life
Data Center Efficiency center design alone. Innovations practices utilized. By optimizing
Data center design—the way facilities such as modular container design, hardware selection, management
are physically constructed, equipped cooling that relies on outside air or and disposal, cloud providers can
with IT and supporting infrastructure, water evaporation, or advanced power outperform on-premise IT in terms
and managed—has a major impact on management through power supply of environmental impact.
the energy use for a given amount of optimization, are all approaches that
• Cloud providers are more likely
computing power. A common measure have significantly improved power
to take an active role in tailoring
of how efficiently a data center usage effectiveness in data centers.
hardware components to the
uses its power is called power usage As cloud computing gains broader specific needs of the services they
effectiveness ratio (PUE). adoption and the share of data run. By collaborating with suppliers
Power usage effectiveness is defined processing performed by modern on specification and design of
as the ratio of overall power drawn data center facilities increases, servers and other equipment for
by the data center facility to the the industry’s PUE averages should maximum efficiency, they realize
power delivered to IT hardware.12 For improve. In parallel, new data center benefits that are too complex for
example, a power usage effectiveness designs continue to push the envelope most corporate IT departments to
of 1.5 means that for every 1 kWh of on driving greater efficiencies. address.
energy consumed by IT hardware, the These two trends will drive greater
• Application code and configuration
data center must draw 1.5 kWh of efficiency in data centers.
provide additional opportunities
energy, with 0.5 kWh used for cooling
for efficiency gains—with cloud
of IT equipment, transforming and
providers more likely to take
conditioning the grid power, lighting
advantage of them. Developers
and other non-IT uses. Standardizing
can write applications with more
and measuring average power usage
efficient processing, memory
effectiveness across companies
utilization and data fetches,
can be difficult. However, the US
ultimately resulting in additional
Environmental Protection Agency
savings of physical consumption
of central processing unit (CPU),
storage, memory and network.
The result is that less physical
infrastructure is needed to deliver
a given application workload.
8

10. Generally speaking, cloud computing from low-carbon electricity sources,
drives the efficiency of IT with such as hydropower or wind energy.
unprecedented economies of scale, This can be accomplished by selecting
higher sophistication and strong a site in a utility region with a lower
incentives to create ongoing carbon emission factor or by actively
efficiencies and continuous purchasing or generating renewable
improvement. Cloud providers spend electricity.
a significant share of their company’s
The difference in location can have a
operational expense on IT—much more
significant impact. For example, data
than an average corporation with its
centers in the US Northwest (where
own IT department. This circumstance
hydroelectric generation is common)
leads to an increased focus on cost
run on power with roughly half the
and efficiency improvement, driving
carbon intensity of the electricity that
optimization of data center and
powers data centers in the Midwest
application performance beyond
(where coal power is common).16 For a
what many businesses can achieve
large data center with 50,000 servers,
on their own. Leading cloud providers
the difference can be equivalent to
will simultaneously address energy
the carbon emissions from thousands
consumption in a variety of ways,
of cars.
whether through application code
optimization, data center temperature Managing carbon intensity through
management, server decommissioning data center location and power
policies or other previously described sourcing strategies, in addition to
approaches. improving energy efficiency associated
with running cloud applications,
Apart from efficiency improvements,
gives cloud providers a powerful lever
cloud providers as well as corporate
to further minimize their carbon
IT departments can reduce carbon
footprint.
emissions by powering data centers
9

11. Case Study – Global Consumer Goods Company
To further validate the findings produced in and Europe already benefits from major
this study about the reduced energy use and economies of scale, with emissions per user far
environmental impact of cloud computing, lower than any small or mid-size deployment.
Accenture and WSP Environment & Energy However, the analysis revealed that moving to
worked with a global consumer goods a cloud solution would save energy and further
company to estimate the potential for reduce carbon emissions by 32 percent.
an improved carbon footprint after moving
The results are in line with the predicted carbon
email and messaging services from its current
savings for large deployments (30 to 60 percent
data centers to Microsoft’s cloud-based
reduction) and help confirm the findings of the
Exchange Online offering.
primary research study. The cloud maintains
The analysis focused on the company’s a strong efficiency margin over on-premise
operations in North America and Europe. solutions, even for efficiently operated, large-
The current large on-premise deployment scale deployments similar to that of this global
of roughly 50,000 users in North America company.
Comparison of On-Premise and Cloud Deployment
for a Global Consumer Goods Company
Microsoft Exchange
On-premise vs. Cloud Comparison,
CO2e per user
= estimated decrease
32% with Microsoft Cloud
On-Premise Microsoft Cloud
10

12. Conclusion & Outlook
Expanding the Cloud companies’ applications, overall aggregate IT efficiency advantages
What will be the environmental net emissions will decrease when in one stroke, instead of investing
impact if cloud-based solutions customers replace existing on-premise in gradual improvements of on-
are widely adopted by businesses servers with cloud services. Thus, premise infrastructure over time.
to replace current on-premise organizations that plan to reduce
• At the data center level, cloud
deployments? energy use and improve their carbon
computing’s growth and drive
footprint can consider migrating
To illustrate the findings from this toward consolidation and
to the cloud as an important
study in an example, it is possible to industrialization will pave the way
means for improving industry-wide
estimate the potential carbon savings for further scale and efficiency.
environmental sustainability.
assuming all US companies with • At the application development
100 to 10,000 employees were using Further Improvements level, software engineers will be
Microsoft Exchange and would switch This study’s finding that companies challenged to code more efficient
from on-premise email servers to the can reduce their carbon emissions applications.
corresponding cloud solution.17 For by 30 to 90 percent by switching
this scenario, the reduction of carbon to a cloud infrastructure is certainly As the efficiency of cloud computing
emissions would be equivalent to the impressive. As impressive as these increases, more services will develop
emissions saved from permanently numbers are, the cloud’s efficiency and while each service or transaction
removing about 100,000 passenger is likely to improve even more over will continue to use less energy,
cars from the road.18 time. Cloud computing is rapidly there is a strong possibility that, in
expanding; demand is increasing and aggregate, computing will use more
This number represents the impact of energy over time. The challenge is
providers are ramping up extra servers
just a single business application for a to ensure that the services provided
to meet predicted future capacity
third of the total US workforce. When in the cloud actually replace current
requirements. As more customers
considered on a global scale and activities of higher carbon intensity.
become cloud users, greater
across an entire spectrum of business As an analogy, a study on music
economies of scale will be reached
applications, the potential impact is distribution shifting to an online
and cloud providers will be able to
very impressive. model demonstrated significant
more accurately predict capacity
These gains will be accelerated by for computing demand. carbon savings—as long as consumers
both Platform-as-a-Service (PaaS) do not also burn the downloaded
As Microsoft and other providers build music onto CDs.19
and Infrastructure-as-a-Service
more data centers based on leading-
(IaaS) offerings, Microsoft’s Windows Cloud computing has enormous
edge designs, and retrofit older data
Azure® services providing an example potential to transform the world
centers, average PUE will continue
for both. As public clouds, such of IT—reducing costs, improving
to improve and per-user footprint of
services allow any IT department efficiency and business agility, and
cloud business applications will shrink
or independent software vendor to contributing to a more sustainable
further over time.
develop cloud-based applications world. This study confirms that
and run them on highly efficient Current technology and practices cloud computing can reduce carbon
infrastructures. The entire IT industry, will also evolve as the data center emissions by 30 to 90 percent for
including users and providers, and cloud market develops further. major business applications today and
will thus be able to reduce its The following trends are likely: that future energy savings are likely
environmental impact through cloud as cloud computing continues to
• At the macro-economic level,
computing. Alternative architectures evolve. Companies who adopt cloud
cloud computing will help achieve
to large-scale public clouds—including computing will accrue the inherent
economies of scale by centralizing
private clouds, community clouds business benefits of this technology,
computing power and improving
and hybrid architectures—can all be and will also play a crucial role
access to variable capacity at a
expected to yield efficiency gains of in making IT more sustainable
more affordable cost.
varying dimensions. by significantly reducing energy
• At the corporate IT leadership
Although the carbon emissions of consumption.
level, moving to the cloud will
cloud providers will increase as they
allow a company to benefit from
run a growing percentage of other
11

13. Appendix
Model Overview • Secondary server consumption data Related Environmental Research.
• The study quantified Microsoft was derived from industry averages CEC-500-2005-110-F. August
Exchange, Microsoft SharePoint and based on J. G. Koomey, “Estimating 5, 2005. And from Christopher
Microsoft Dynamics CRM application Total Power Consumption by R. Hannemann, Van P. Carey,
fulfillment activities for both cloud Servers in the U.S. and the World” – Amip J. Shah, Chandrakant Patel,
and on-premise scenarios by dividing February 15, 2007; and GreenGrid. "Lifetime energy consumption of
the total energy consumption and an enterprise server," ISEE, pp.1-5,
• Time period considered: a one-year
resulting carbon footprint against the 2008 IEEE International Symposium
application licensing or subscription
number of active users for a on Electronics and the Environment,
agreement.
given application. 2008.
• GHG emissions (“carbon
• The Microsoft Dynamics CRM results emissions”) included are stated Process Energy for
in Figure 1 are based as carbon dioxide equivalent IT Infrastructure
on the planned cloud installation (CO2e) emissions and take into • Power consumption of Microsoft’s
operating at target capacity. account the six primary GHG gases servers was based on direct power
• The model was independently including, CO2 (carbon dioxide), measurement of application-specific
developed based on ISO 14044 SF6 (sulphur hexafluoride), CH4 server racks from Microsoft data
guidelines for Life Cycle Assessment, (methane), N2O (nitrous oxide), centers.
BSI PAS 2050 Specifications for HFCs (hydrofluorocarbons), PFCs • Estimated power consumption
the Assessment of Greenhouse (perfluorocarbons). of servers within the on-premise
gas (GHG) Emissions of Goods • The study includes the use phase of environment was based on
and Services, and the WRI/WBCSD the product by the customer. While industry-standard figures provided
GHG Protocol. use is assumed to be the same by Hewlett Packard and verified
• The study and related analytical rate for cloud and on-premise, the by Accenture and Avanade using
modeling builds upon the work efficiency and energy consumption specific server configuration sizing
previously completed by Accenture associated with the two scenarios calculations. A mixture of different
and WSP Environment & Energy to are different. vendors' on-premise systems was
assess the carbon impacts of Digital assessed, rather than any single
Materials server product.
Distribution for Microsoft Volume
• Primary materials included in the
Licensing of Microsoft Office. • The model includes essential power
study consisted of servers and
• The aggregated results in this report related network equipment used for critical IT environment and
have been calculated based on a to host an application. Embodied utilizes a Microsoft data center-
scope limited to North American emissions from physical hardware specific power usage effectiveness
and European regions. Customers were estimated based on the (PUE) ratio and an organization-
operating in different regions will be weight and composition of each type specific on-premise PUE ratio
subject to different carbon emission component. based upon EPA, Green Grid and J.G.
factors and specific data center Koomey research.
• Embodied emissions from physical
utilization rates that could affect the • A storage consumption and network
infrastructure included servers, but
findings of a similar study. usage efficiency ratio was also used,
not facilities and other equipment.
• Primary data was provided by based upon Microsoft data and
• Emissions related to the material EPA, Green Grid and J. G. Koomey
Microsoft, Accenture and Avanade
manufacture, assembly and research, validated by Accenture
(using actual measurements or
recovery of servers and networking and Avanade.
conservative estimates) for cloud and
equipment are based on a 3.5-
on-premise scenarios. • For on-premise scenarios, a server
year refresh rate for data center
• Secondary data for materials hardware. Life Cycle Inventory of redundancy factor of 2X was
was derived from the EcoInvent a server derived from Masanet E., assumed. For cloud, the model relies
database and other publicly available et al. ‘Optimization of Product Life on Microsoft’s actual server counts
databases collated Cycles to Reduce Greenhouse Gas which include redundancies to meet
in SimaPro. Emissions in California.’ California corresponding service levels.
Energy Commission, PIER Energy-
12

14. Supply Chain Logistics & Model Exclusions Microsoft Project Sponsorship:
Distribution The model used in the research study Rob Bernard, Chief Environmental
• Emission factors for transportation excluded the following factors: Strategist
were derived from the WRI/WBCSD • Energy consumed during software Francois Ajenstat, Senior Director
GHG Protocol CO2 emissions from development. Environmental Sustainability
Mobile Sources.
• Tertiary suppliers and process Mark Aggar, Environmental
• Emissions were calculated based materials that are not significant Technologist
on frequency, modes, distance and (i.e., that do not constitute an input
weight (ton-kilometers) of the to 95 percent of the product or Accenture Authors and Key
hardware. process). Contributors:
• Servers are assumed to be • Offsetting of emissions from any Dave Abood, Managing Director,
manufactured in Asia and other part of the supply chain. Sustainability Services NA
transported by marine freight
to North America or Europe. • Embodied energy of capital Robin Murdoch, Senior Executive,
equipment, transportation vehicles, Cloud Strategy
End-of-Life (EoL) Processes buildings and their energy use Stephane N’Diaye, Senior Manager,
• End-of-life calculations include the not directly related to servers and Sustainability Services NA
emissions associated with recycling associated equipment.
and land filling IT equipment David Albano, Manager, Management
• Maintenance of capital equipment. Consulting, Sustainability
amortized over the useful life
of the equipment. • Refrigerants (except where used Andri Kofmehl, Manager, Management
in the primary production of raw Consulting, Sustainability
• The study used a conservative
inputs).
assumption of 20 percent recycling Teresa Tung, Manager, Accenture
and recovery for servers and Technology Labs
network equipment.
Ari van Schilfgaarde, Consultant,
Management Consulting, Sustainability
WSP Environment & Energy
Authors and Key Contributors:
Andrew Armstrong, Vice President
Josh Whitney, Senior Project Director
13

15. 1 “On-premise" refers to running an IT 8 Energy use to transmit data between users 11 The Green Grid. Five Ways to Reduce Data
application on an organization’s own servers, and servers was not modeled in detail. For the Center Server Power Consumption. 2008.
either in an office space or a corporate data type of applications analyzed in this study, 12 The Green Grid. Green Grid Data Center
center rather than on a shared cloud or hosting network equipment typically consumes around
Power Efficiency Metrics: PUE and DCiE. 2008.
infrastructure provided by another company. ten times less energy than servers and is
thus not considered a very significant factor. 13 US Environmental Protection Agency (EPA).
2 Microsoft’s cloud-based Exchange and
The greater distances data travels in a cloud ENERGY STAR Data Center Infrastructure Rating
SharePoint offerings (Exchange Online and
scenario may be more than offset by low Development Update. 2009.
SharePoint Online) are both a part of the
utilization of corporate network equipment or 14 US Environmental Protection Agency (EPA).
Business Productivity Online Standard Suite
employees accessing corporate servers remotely.
(BPOS) and are sold either as a bundled Report to Congress on Server and Data Center
However, when analyzing the environmental
business solution or individually, depending Energy Efficiency. 2007.
footprint of data-intensive (consumer)
on customer preference. Specific on-premise 15 Microsoft. Microsoft’s Top 10 Business
applications, such as music download or
versions analyzed were Microsoft Exchange
video streaming, data transfer contributes Practices for Environmentally Sustainable
2007, Microsoft SharePoint 2007, and
a significant share to the overall footprint Data Centers. 2009.
Microsoft Dynamics CRM 4.0.
and requires more in-depth analysis. 16 eGrid. GHG Annual Output Emission Rates.
3 Accenture and UN Global Compact. A New
9 Microsoft’s cloud-based Exchange and 2008.
Era of Sustainability. 2010.
SharePoint offerings (Exchange Online and 17 Assuming that the number of email
4 Global e-Sustainability Initiative (GeSI). SharePoint Online) are both a part of the
accounts is roughly the same as the number of
SMART 2020: Enabling the Low Carbon Business Productivity Online Standard Suite
employees and that companies either retire or
Economy in the Information Age. 2008. (BPOS) and are sold either as a bundled
re-allocate the on-premise server capacity that
5 Microsoft Global Foundation Services. business solution or individually, depending
was dedicated to email, calendar and contacts.
on customer preference. Specific on-premise
A Holistic Approach to Energy Efficiency Using employee numbers from US Census data.
versions analyzed were Microsoft Exchange
in Data Centers. 2010. 18 According to the US EPA Greenhouse Gas
2007, Microsoft SharePoint 2007, and
6 Accenture and WSP. Demonstrating the Microsoft Dynamics CRM 4.0. Equivalencies Calculator.
Benefits of Electronic Software Distribution. 10 Silicon Valley Leadership Group. Data Center 19 Weber, Koomey, Matthews. The Energy and
2009.
Energy Forecast. 2008. Climate Change Impacts of Different Music
7 Gloabal e-Sustainability Initiative (GeSI). IDC. The Economics of Virtualization: Moving Delivery Methods. 2009.
An Assessment Methodology. 2010. Toward an Application-Based Cost Model.
2009.” – 14

16. About Accenture About WSP Environment & About Microsoft
Accenture is a global management Energy Founded in 1975, Microsoft is the
consulting, technology services WSP Environment & Energy worldwide leader in software, services
and outsourcing company, with is one of the world’s leading and solutions that help people and
approximately 204,000 employees global consultancies, delivering businesses realize their full potential.
serving clients in more than 120 solutions to environmental, With 89,000 employees across
countries. Combining unparalleled energy, and sustainability issues. its business divisions and global
experience, comprehensive capabilities With over 1,000 people across 65 subsidiaries, the company generated
across all industries and business offices globally, WSP Environment revenues of US$ 62.5 billion in
functions, and extensive research & Energy offers a full-service fiscal year 2010. Its home page
on the world’s most successful consulting practice to a range of is www.microsoft.com.
companies, Accenture collaborates commercial and industrial clients,
with clients to help them become many of whom are Fortune 500
high-performance businesses and and FTSE 250 companies. WSP
governments. The company generated helps its clients increase business
net revenues of US$21.6 billion for the performance through process
fiscal year ended Aug. 31, 2010. Its improvement, risk mitigation,
home page is www.accenture.com. and design and implementation
of sustainable business practices.
WSP Environment & Energy is part
of WSP Group plc. Its home page is
www.wspenvironmental.com/usa.
Copyright © 2010 Accenture
All rights reserved.
Accenture, its logo, and
High Performance Delivered
are trademarks of Accenture.
This document makes reference
to trademarks that may be
owned by others. The use of such
trademarks herein is not an assertion
of ownership of such trademarks
by Accenture and is not intended
to represent or imply the existence
of an association between Accenture
and the lawful owners of such
trademarks.
Copyright © 2010 WSP
Environment & Energy.