CLOUD COMPUTING: RETURN ON INVESTMENT
The Portuguese Higher Education Case Study
Jorge Sousa
2
, Pedro Assis
1,2
and Miguel Leitão
1,2
1
School of Engineering, Polytecnic of Porto, Rua Dr. António Benardino de Almeida, 431, Porto, Portugal
2
EuroCloud Portugal, Rua Dr. António Benardino de Almeida, 431, Porto, Portugal
Keywords: Cloud Computing, ROI, Survey, Indicators, Metrics, Population and Sample.
Abstract: This work is about return on investment (ROI) estimation based on a set of scenarios related with Cloud
services adoption by Portuguese higher education (HE) institutions. The adopted methodology required the
development of a survey and its distribution among institutions. The collected data allowed us to evaluate a
set of indicators and metrics in order to design ROI models. With such models it was possible to estimate
the cost benefit of Cloud Computing paradigm in the context of Portuguese HE infrastructures and services.
1 INTRODUCTION
Cloud Computing is undoubtedly in vogue. The
promise of productivity increase and cost reductions
is leading to a growing interest in assessing Cloud
services. However, much work is still to be done
until useful evaluation tools and methodologies are
agreed upon. Up to now, studies addressing the
return on Cloud Computing investments are
inconclusive, revealing that this research is still in its
infancy. Nonetheless, we found several studies that
are valuable contributions, namely (Harms,
2010),
(Linthicum, 2011), (Mayo, 2009), (Misra, 2011) and
(The Open Group, 2010).
This paper describes a case study on the
assessment of Cloud Computing return on
investment (ROI) in the context of the Portuguese
higher education system. The proposed methodology
is based on the development of a survey distributed
among a sample of Portuguese higher education
institutions (HEI). The survey results were
assembled to identify a set of indicators and metrics
that are relevant to the quantification of benefits and
costs associated with the use of this paradigm.
Cloud Computing assessment is not only about
direct economic advantages, but also about
intangible benefits. Traditionally such benefits are
not quantifiable; nevertheless they reflect added
value to business. Therefore, Cloud ROI analysis
was split into financial and non-financial.
Financial benefits are those that are measurable.
These were categorized into five major groups:
Hardware, Software, “Human Resources”, “Energy
Consumption” and “Data Center Space”. Non-
financial benefits include a large and diverse set of
aspects for which it is difficult to establish the
principle of “direct causality.” Such factors are
evaluated differently depending on the agent
subjective appreciation. These factors were grouped
here under Productivity and “Systems
Administration.” Finally, there is the “Automatic
Provision” area. This area doesn’t fit in either ROIs,
but has impacts in both of them.
2 SURVEY
The survey was developed aiming the evaluation of
a latent variable, in this case the return on
investment. The term “latent variable” is used to
represent a variable that cannot be observed or
measured directly, but can be inferred from a
coherent set of other variables, which can be
observed or measured. To this end, a Goggle Docs
questionnaire of 20 to 26 (depending on the answers
given) closed ended questions was distributed
among institutions’ network and information
systems managers.
The sequence of questions guides the respondent
through three major groups (scope) of subjects. The
first group relates with the characterization of the
institutions from the ICT point of view. It contains
questions like “the number of ICT staff” and “the
number of servers installed in the data center.” The
472
Sousa J., Assis P. and Leitão M..
CLOUD COMPUTING: RETURN ON INVESTMENT - The Portuguese Higher Education Case Study.
DOI: 10.5220/0003925404720475
In Proceedings of the 2nd International Conference on Cloud Computing and Services Science (CLOSER-2012), pages 472-475
ISBN: 978-989-8565-05-1
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
second group concerns the knowledge about Cloud
Computing and its associated technologies: “do you
make use of virtualization?” and “do you use an
external webmail service like Gmail?” The third
group is about the way people are using Cloud. This
group is further classified into three sub-groups,
reflecting the degree of Cloud adoption. Examples
of questions in this group are: “why was this
paradigm adopted?” and “what were your fears?”
3 POPULATION AND SAMPLE
The target population was the set of all the 410
institutions of the Portuguese higher education
system. Taking into consideration such high number
of institutions, their geographical location and the
lack of resources of this research, conducting a
census was not a viable option. It was decided to
work with a sample of institutions statistically
representative (≥30) to allow statistical
inferences about the population. Therefore, a sample
of 43 institutions, 10.5% of the population,
providing a maximum error (ε) of 11%, was built
based on the concept of stratified sampling. This
sampling technique is a probability method such that
sub-populations are included in a balanced way.
Four levels of stratification were used based on the
following criteria: geographical location, education
subsystem, size and fields of studies. Completed the
stratification process, a simple random sample
technique was applied to each group.
4 SURVEY RESULTS
According to this study the percentage of the
Portuguese HEIs that use Cloud services is 18.6%
(only 8 of the 43 surveyed institutions). The majority
of these institutions are using private Cloud model
from the provider point of view (i.e., supporting
institutions’ services). Based on this result, it is
inferred, with a 95% confidence interval, that 7.6%
to 29.6% of the whole Portuguese HEI population
are using Cloud Computing.
The remaining 35 surveyed institutions do not
use Cloud services. Most of these institutions are
interested in the hybrid Cloud model either from the
consumer or provider points of view. This subset
was asked about this Cloud adoption in the future,
which 42.9% said “yes.” In another words, 15 of
these 35 institutions are considering the Cloud
Computing adoption. However, based on that result
it is not possible to infer about the Portuguese HEIs
(whole population) that are evaluating the use of
Cloud services. The statistical analysis made,
namely the binomial test, was inconclusive.
4.1 Institutions that use Cloud
Computing
Two main reasons were reported for Cloud adoption:
the expectation of cost reductions (75% of the
institutions of this subset), and the flexibility of
sharing resources (50%). In regard to concerns, 75%
of the institutions said their biggest fear came from
provider lock-in. This concern relates to the
difficulty in migrating between service providers
due to many barriers, namely the lack of information
portability. Security problems and dependence on an
external service provider appeared in second place
for 50% of the Portuguese HEIs of this group.
The 8 Portuguese HEIs that are using Cloud
services were asked about the impact that such
adoption had on their ICT investment profile: 35.7%
answered that it decreased by 10%. But, similar
number of respondents said that no significant
variation occurred. One should note that 25% of the
answers indicate an investment growth. This might
reveal that institutions have on site both kinds of
infrastructures: traditional ICT and Cloud.
When asked about cost reductions attained in a
number of key areas, like staff training, systems
administrators and server consolidation, the majority
of the surveyed institutions reported revenues as low
as 10%. Based on the answers given it was possible
to identify, by a modal scores analysis, a set of
indicators (“Server consolidation”, “Electricity
consumption reduction”, “Operating systems
licenses reduction” and “Data center footprint
reduction”) that will be used to estimate the financial
ROI. The average cost reductions reported by these
indicators were between 2.5 and 12.5%. A similar
analysis was made to identify the set of indicators
(“Increase productivity and efficiency” and
“Increase use of resources”) for the non-financial
ROI evaluation. Again, the gains of these indicators
vary between 2.5 and 15%. Finally, in the area of
automated provision these institutions reported an
average gain of 5% associated with the decrease of
provision time.
4.2 Institutions that are Evaluating
Cloud Computing Adoption
The 15 institutions, which do not use Cloud services,
but are evaluating its adoption, were questioned
about the benefits associated with the use of Cloud
CLOUDCOMPUTING:RETURNONINVESTMENT-ThePortugueseHigherEducationCaseStudy
473
Computing. Most of the institutions (66.7%) are
expecting to reduce costs of ICT investment.
However, 60% of the choices fall on the availability
and ubiquity of the service. Note that the
maintenance responsibility transfer from site to the
service providers is not among the benefits that the
institutions mostly associate with this paradigm.
When asked about their greatest concerns, 73.3%
reported security problems as the main barrier to
adopt Cloud services. Also, 60% of the institutions
of this group are reluctant to become dependent on a
single service provider (external) and 46.7% report
the lack of control over the resources as a drawback
in its adoption. The lack of interoperability between
service providers is not a top concern of these
institutions, meeting only 13.3% of indications.
In regard to the expectations on ICT cost
reductions after Cloud Computing adoption, one
third of the institutions reported a reduction
expectation between 10 and 30%. The same number
of respondents expects a greater gain, between 30
and 50%. On the contrary, 20% of the respondents
do not expect any cost reductions. To attain more
detailed information concerning this issue, it was
further asked to specify the expected investment
reduction for the same set of key areas as the
previous group (Section 4.1). In this case the
institutions reveal higher expectations as they report
cost reductions for almost all areas between 10 and
40%. A single area was left out: the training area.
5 ROI FORMULATION
Return on investment (ROI) is a financial
measurement used to evaluate the efficiency of an
investment. ROI is a popular metric because of its
versatility and ease of use. A common formulation
for ROI is given by (1), where “Gain from
investment” refers to the profit obtained from the
investment made. Based on this definition if an
investment does not have a positive ROI then it
should not be undertaken.
=
−
(1)
Given the nature of the surveyed organizations
the concept of gain, from an ICT investment, does
not have a straight forward interpretation: It is not
within the HEI purpose the trading of ICT services,
but these services are required as they allow
institutions to pursue their goals. In this context, and
taking into consideration that the selected indicators
(both financial and non-financial) reflect investment
cost reductions due to Cloud Computing adoption,
(1) must be rewritten.
We start by expressing ICT ROI (2) and Cloud
Computing ROI (3) using (1). Assuming that Cloud
Computing investment gain relates to the investment
gain of a non-cloud ICT scenario through a linear
relationship (4) parameterized by α and β, where α
reveals the relative evolution of the two gains over
time, and β reveals the added value of adopting
Cloud Computing, both ROIs can be related as
described by (5).
=
−
(2)
=
−
(3)
=
×+
(4)
=
+1
×
×+
−1
(5)
Considering that the variation of investment gain
is given by (6) and a somewhat conservative
approach of the Cloud Computing gain is adopted, as
no gains are associated with its adoption (the gain is
the same as for the ICT use case, i.e.
=1 and
=0), only costs reduction contributions, then it is
possible to express Cloud Computing ROI based on
the ICT ROI (prior to Cloud support) and the
investment cost reduction (7).
∆
%
=
−
(6)
=
+∆
%
1−∆
%
(7)
6 RESULTS REVIEW
Expression (7) limits this study to a single indicator
analysis, meaning that the impact of each indicator
was assessed individually. A multi-indicator analysis
was not conducted due to the lack of mathematical
models to describe such dependencies, although we
have found evidences of such behavior between
indicators.
Distinct financial evaluations were performed
based on the answers received from the first two
groups of institutions: those who already use Cloud
Computing and those who are evaluating its use. In
both cases the data collected represent investment
cost reductions (6) for a set of financial indicators. A
comparison between the evolutions of the expected
CLOSER2012-2ndInternationalConferenceonCloudComputingandServicesScience
474
ROI for both groups of institutions shows a large
discrepancy. While the institutions that are
considering using cloud services reveal a significant
cost savings expectation, the institutions that
actually use such services argue that these reductions
exist, but are moderate.
For the indicator “Reduction in the number of
licenses for operating systems,” it was considered
that there was a direct relation between the reduction
in the number of licenses and its costs. However, we
may have an additional gain due to variations in
licenses’ costs. In order to explore such possibility,
let’s consider N as the number of licenses prior to
the adoption of Cloud Computing and
the cost
associated with those licenses. With the use of Cloud
services the number of licenses is now M and its
associated cost
ρ
⁄
. In a scenario of multiple
operating systems, ρ measures the relative evolution
of the two costs due to the different reductions
achieved for each type of licenses. Thus, the new
cost is given by (8).
=
M
N
×
×ρ
(8)
Assuming the variation in the number of licenses
is given by (9), equation (8) can now be rewritten as
(10).
∆
=
N−M
N
(9)
=
1−∆
×
×ρ
(10)
Using (6) we can now express ∆
%
as a
function of ∆
and ρ (11). Finally, the Cloud
Computing ROI is expressed, for this indicator, by
equation (12).
∆
%
=1−ρ
1−∆
(11)
=
+1−ρ
1−∆
ρ
1−∆
(12)
Expression (12) reveals a reduction of costs
parameterized by ρ. Three scenarios can be
envisaged based on the weight cost of the licenses
fees that are no longer paid compared to the total. In
the first scenario, such reduction is similar to just
considering the reduction of the total number of
licenses, i.e. the weight of the licenses that have
been discontinued is not significant compared to the
total (e.g. =0.9). In the second case, the reduction
achieved is greater than the previous scenario
because the weight of the fees is important to bear
against the total (e.g. =0.6). In the third scenario,
the most favorable towards Cloud, due to the
significant weight of the cost of the licenses that are
no longer paid (e.g. =0.3). A similar analysis
applies to the “Systems Administrators” indicator.
7 CONCLUSIONS
A glimpse of the status of Cloud adoption by the
Portuguese HEIs is provided by this work. Also,
Cloud ROI models were elaborated to estimate,
based on the collected data, the return on Cloud
investment, but only for a single indicator analysis.
Such data and results will be refined and compared
with other sources as they are made available. Also,
further ROI models will be developed to address
Cloud lifecycle. Such work will target specific use
cases, at different Cloud layers, focusing near-term
versus long-term analysis.
ACKNOWLEDGMENTS
The authors would like to thank Inês Dutra and
Sandra Ramos for all valuable comments and
revisions. This publication was partially supported
by a grant from the EuroCloud Portugal Association.
REFERENCES
Harms, R., Yamartino, M, 2010. The Economics of the
Cloud. Microsoft.
Linthicum, D., 2011. ROI of the Cloud Part 1: PaaS and
Part 2: IaaS. Microsoft.
Mayo, R., Perng, C., 2009. Cloud Computing Payback: An
explanation of where the ROI comes from. IBM.
Misra, S., Mondal, A., 2011. Identification of a company’s
suitability for the adoption of cloud computing and
modelling its corresponding Return on Investment,
Mathematical and Computer Modelling, Vol.53,
Issues 3–4, pp 504-521. Elselvier.
The Open Group, 2010. Building Return on Investment
from Cloud Computing. Document No.: W104
CLOUDCOMPUTING:RETURNONINVESTMENT-ThePortugueseHigherEducationCaseStudy
475
