Content uploaded by Vinicius Cardoso Garcia
Author content
All content in this area was uploaded by Vinicius Cardoso Garcia
Content may be subject to copyright.
Accounting Models for Cloud Computing: A
Systematic Mapping Study
Francisco Airton Pereira da Silva1, Paulo Anselmo da Mota Silveira Neto1,
Vinicius Cardoso Garcia1, Rodrigo Elia Assad1, Fernando Antonio Mota Trinta2
1Federal University of Pernambuco (UFPE) – Recife, Pernambuco, Brazil
2Federal University of Cear´
a (UFC) – Fortaleza, Cear´
a, Brazil
{faps,pamsn,vcg,rea}@cin.ufpe.br {fernando.trinta}@virtual.ufc.br
Abstract—Cloud services change the economics of computing
by enabling users to pay only for the capacity that they actually
use. In this context, cloud providers have their own accounting
models including their billing mechanisms and pricing schemes
to achieve this efficient pay-as-you-go model. Thus it is important
to study this heterogeneity aiming to map out the existing
accounting models to become possible new proposals or future
standardizations. Therefore, this paper focuses on mapping
accounting models for cloud computing, where a mapping study
process was undertaken, and a total of 23 primary studies were
considered, which evidenced 5 accounting models, 23 different
pricing scheme types and 4 primary studies related to SLA
(Service-Level Agreement) composition. Although the significant
number of studies found address grid computing it was possible
to identify one accounting model which was very complete from
different points of view for cloud environments.
Index Terms—Cloud Computing; mapping study; pricing
scheme; accounting model; Service Level Agreement.
I. INTRODUCTION
Cloud computing has become an established paradigm for
running services on external infrastructure, where virtually
unlimited capacity can be dynamically allocated. However this
unlimited aspect in some cases can become expensive, and
research projects have tried to mitigate it through the devel-
opment of new architectures, exploring different accounting
models [1] [2] [3] [4].
Accounting in cloud computing is a recent discipline, hence
there have few attempts to find a model which considers all the
accounting requirements, and none work has tried to address a
mapping of the existing accounting models that could identify
research gaps and encourage future proposals.
In this context, this paper introduces a mapping study
performed between July and December, 2011, addressing
accounting models for Cloud Computing environments and
other aspects related also to Grid Computing.
We had to encompass the grid computing research field,
mainly due to three considerations. The first point is the
correlated aspects between cloud and grid computing, the
second point is the older grid origin with probable relevant
contributions and as final reason, the existing mature account-
ing models under this research area.
In [32] the authors perform a comparison between the six
most known accounting systems in grid computing, evidencing
the advantages and disadvantages of them whereas allowing
to realise what aspects they have in common.
First, they use a proper taxonomy to describe their functions
which make part of an accounting process (a set of operations
that manages the data regarding the use of the resources [5]).
Next, they present a measurement unit mechanism to apply
under the resource consumption and accordingly charge for it,
called pricing scheme [6].
Finally, all of them worry about QoS Requirements and
explores how to monitor this Quality of Service. In some cases
establishing Service Level Agreements (SLA).
Based on aforementioned items and previous literature
investigation, four research questions were derived to guide
this mapping study, as follows:
•RQ1: Is there any taxonomy for concepts related to
accounting process in cloud computing?
•RQ2: What are the existing accounting models for cloud
computing?
•RQ3: What are the existing pricing schemes for cloud or
grid computing?
•RQ4: What are the aspects taken into account to compose
a SLA in cloud/grid computing scenario?
The remainder of the paper is structured as follows: Sec-
tion II introduces the related work; Section III presents the
systematic mapping study process; Section IV describes the
main findings of the study; Section V presents the analysis
of the results, studies classification and mapping; Section
VI introduces some threats to validity. Finally, Section VII
presents the conclusions and future research.
II. RELATED WORK
Basically our research started motivated by the evolution in
federated cloud infrastructures field, which two works stands
out (RESERVOIR and JiT Clouds).
RESERVOIR Project [2] presents an architecture (including
an advanced accounting model) that allows providers of cloud
infrastructures to dynamically partner with each other to create
a virtually infinite pool of resources.
JiT Clouds Project [7] also allows providers of cloud
infrastructures to dynamically partner with each other, but with
the advantage where providers does not need keep dedicated
Int'l Conf. Grid Computing and Applications | GCA'12 |
3
resources to meet the service providers demands, however does
not have an accounting model.
In [8] the authors present a comparative review of grid and
cloud computing pricing models. Unlike our proposal, this
paper is not a systematic study and related only with our RQ3.
III. SYSTEMATIC MAPPING STUDY PROCES S
A Mapping Study is a systematic process that provides an
overview and summarizes published paper results of a partic-
ular research area, by answering questions and categorizing
the studies. As main benefit, it can be used to identify gaps
in the existing research that will lead to topics for further
investigation [9].
Therefore, a Systematic Mapping Study was used in this
research to “map out” the accounting models for cloud com-
puting, performing five steps (Questions Definition, Search,
Screening, Keywording and Extraction) [9].
A. Conduct Search
The strategy used to construct the search terms, follows
the same approach used in [10], since it is systematized in
essence and defines steps to derive the search strings from
the questions and the viewpoints of experts in the area and
relevant papers. The strategy steps are described as follows:
•Derive major terms from the questions by identifying the
population, intervention, outcomes and study design;
•Identify, by inquiries with experts in the field, alternative
spellings and synonyms for major terms; and
•Check the keywords in the relevant papers.
The complete list of search strings and their combination
are presented in Table I.
TABLE I
SEARCH STRING
SLA OR “Service Level Agreement” OR billing OR pricing
OR payment OR accounting AND “cloud computing” OR
“grid computing” OR “Infrastructure as a Service” OR “Plat-
form as a Service” OR “Software as a Service”
Firstly an automatic search was conducted in different
search engines (IEEEXplore, ACM Digital Library, Scopus
and ScienceDirect digital databases). It is important to men-
tioned that all search strings were calibrated regarding to
each search engine. Next, a manual search was performed by
visiting some important conference proceedings. As a results
from the application of both search strategies 580 studies were
collected.
At this point, the studies were excluded according to the
exclusion criteria:
•Studies did not address or just mentioned accounting
models/processes, pricing schemes, SLA composition on
cloud/grid computing;
•Studies only available as abstracts or presentations; and
•Duplicate studies. When a study has been published in
more than one publication, the most complete version will
be considered.
B. Screening of Papers
Firstly, the exclusion criteria were applied on the title and
abstract of the identified studies, resulting in 98 studies being
selected. The large number of duplicated studies contributed
to this large difference. Next, a second filter was applied,
analysing the introduction and conclusion, which resulted in
23 studies ([1], [2], [3], [4], [5], [6], [8], [11], [12], [13], [14],
[15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25]
and [26]).
C. Keywording
A classification scheme was built which analysed the ab-
stract, titles and keywords of the selected primary studies to
identify different facets. This way, three different facets were
used. They are described following:
•Contribution Type: Method, Process, Technique, Model
and Framework [27];
•Accounting Model Features: Pricing, Metering, Media-
tion, Accounting, Roaming, Billing, Charging, Financial
Clearing, Cloud Federation, Just in Time Clouds, User
Interface, Security Support, SLA Support and Variable
Payment Models;
•Research Type: Validation Research, Evaluation Re-
search, Solution Proposal, Philosophical Papers, Opinion
Papers, and Experience Papers [28] (see definitions in
Table II).
TABLE II
RESEARCH TYPE FACET [28]
Class Description
Validation Research Techniques investigated are novel and have not yet been
implemented in practice. Techniques used are for example
experiments, i.e., work done in the lab.
Evaluation Research Techniques are implemented in practice and an evaluation
of the technique is conducted. That means, it is shown how
the technique is implemented in practice (solution imple-
mentation) and what are the consequences of the implemen-
tation in terms of benefits and drawbacks (implementation
evaluation). This also includes identification of problems in
industry.
Solution Proposal A solution for a problem is proposed, the solution can
be either novel or a significant extension of an existing
technique. The potential benefits and the applicability of the
solution is shown by a small example or a good line of
argumentation.
Philosophical Papers These papers sketch a new way of looking at existing things
by structuring the field inform of a taxonomy or conceptual
framework.
Opinion Papers These papers express the personal opinion of somebody
whether a certain technique is good or bad, or how things
should been done. They do not rely on related work and
research methodologies.
Experience Papers Experience papers explain what and how something has been
done in practice. It has to be the personal experience of the
author.
D. Data Extraction
A data extraction form was designed in order to gather the
required information to address the objectives of this study,
classifying and answering the research questions. The full
paper was read and the following information was extracted
4
Int'l Conf. Grid Computing and Applications | GCA'12 |
from each study: the research categorization (Contribution
Type, Accounting Model Features and Research Type), in
addition the information required to answer some of the
research questions.
IV. RESULTS
In this section, each topic presents the findings regarding
to a specific research question, highlighting the evidences
gathered from the data extraction process.
A. RQ1 - Is there any taxonomy for concepts related to
accounting process in cloud computing?
In our research only one primary study effectively answered
this question. The study [5] presents a taxonomy of full
accounting process and its functions from the resource usage
to the financial clearing. It is not applied only to cloud
computing, but other areas related to Services on the Internet
(see Table III).
TABLE III
TAXONOMY OF ACCOUNTING PROCESS [5]
Concept Function
Pricing Function of giving a price to a certain resource usage.
Metering Collects raw information regarding the resource usage of a
certain service by a consumer and its usage.
Mediation Is intended to do a first treatment of raw technical data by
transforming these metering records into a data format that
can be used for storing and further processing.
Accounting Has the function of filtering and treat more accurately the
records passed by mediation function.
Roaming Allows using more than one provider while maintaining a
formal, customer-vendor relationship just with one.
Billing Also called of invoicing, is the process of transforming
charge records into the final bill, summarizing the charge
records of a certain time period and indicating the amount
of monetary units to be paid by the customer.
Charging Is the process of calculating the cost of a resource usage,
the function that translates technical values into monetary
units by applying a pricing function to the session records.
Financial
Clearing
Includes activities from a commitment for a transaction to its
settlement. In the case of resource accounting, this function
implies the payment of a bill.
Although it was found only this taxonomy formally defined,
other terms are widely used with the same meanings. For ex-
ample, monitoring has the same sense of metering. According
to [25] the metrics generated by the monitoring function can be
used both for accounting purposes as for performance analysis.
In other study [5], monitoring is a sub-function of metering
that collects the information of a resource usage as raw data
and provides usage metrics to the metering function.
B. RQ2 - What are the existing accounting models for cloud
computing?
When performing the analysis, were found five primary
studies ([1], [2], [3], [4] and [25]) that proposed some kind of
accounting model, summarized following:
a) Flexible Accounting Model [1] - This paper proposes
a flexible accounting model suitable to any service of cloud
computing. This model is based on the accounting process of
the Internet and it can fit any pricing scheme using jBilling
accounting platform and mainly through the use of IPDR
(Internet Protocol Detail Record).
b) A Model for Federated Clouds [2] - This primary study
presents a solution for an accounting and billing architecture
for use in federated cloud environments like the RESERVOIR
project (funded by European Union). The model is organized
in layers(Accounting, Billing and Business Layer)
c) ABS for SOA [3] - This primary study presents a
framework wherein authentication of the clients and billing
of services used by client is carried out. So this paper threats
the security as an essential requirement in billing services. like
generates instances of virtual machines for a particular time
period ordered by user (time-based pricing scheme) in safe
mode.
d) THEMIS [4] - This model proposed a mutually (provider
and user) verifiable billing system called THEMIS to Cloud
Computing scenario in which has as main requirements the
transparency, security and low latency in billing transactions.
Thus, the system introduces the concept of a Cloud Notary
Authority to supervise billing transactions, using a level of
security that is identical to that of a Public Key Infrastruc-
ture (PKI), combating the malicious behaviour of users and
providers.
e) Cloud Supply Chain [25] - This model proposes the
Cloud Supply Chain concept, which represents a network
of interconnected businesses in the cloud computing area
involved in the end-to-end provision of product and aggregated
service packages required by end cloud service customers.
This includes the actual provisioning of infrastructure services
and the Information Model supporting monitoring, accounting
and billing processes.
C. RQ3 - What are the existing pricing schemes for cloud or
grid computing?
Table IV summarizes all the pricing schemes found with
their respective concepts and in which study they were dis-
cussed.
There is a lot of work that mention some type of pricing
scheme. However it is used different terms to the same pricing
scheme meaning. In [8], [15] and [16] the authors refers to
pricing schemes as pricing models and specially in [8] the
pricing models are grouped in a general way called economic
models. For example, the economic model Commodity Market
(price defined based on amount of resource that users used)
has as pricing models: Usage Duration and Flat Fee.
D. RQ4 - What are the aspects taken into account to compose
a SLA in cloud/grid computing scenario?
In order to cloud providers supply clients with services that
meet their quality constraints, they both need to negotiate the
clients requirements and the provider’s infrastructure capabili-
ties. It is known as Service Level Agreement (SLA) . However,
this is not an easy task, according to [29] there are many
difficulties to formalize a SLA, such as lack of flexibility and
precision. This way, to compound a SLA it is important to
know which aspects have to be taken into account.
Int'l Conf. Grid Computing and Applications | GCA'12 |
5
TABLE IV
PRICING SCHEMES
Pricing Scheme Definition Studies
Time-based Pricing based on how long a service is used. [24], [1], [6]
Paris-Metro pricing Used for shared resources. Resources are split by the amount of users per split. [1]
Priority pricing Services are labelled and priced according to their priority. [1]
Flat-rate A fixed tariff for a specified amount of time. [24], [1], [6]
Edge pricing Calculation is done based on the distance between the service and the user. [1]
Responsive pricing Charging is activated only on service congestion. [1]
Effective bandwidth pricing Charging is based on an expected usage function. [1]
Proportional fairness pricing It is according to the user’s willingness to pay, in other words, It is based on the real value of
product or service.
[16], [1]
Cumulus pricing Based on flat pricing and dynamically priced by using a credit point system. [1]
Session-oriented Based on the use given to the session. [1]
One-off charge per service One charge per service session. [1]
Usage-based Pricing based on the general use of the service for a period of time, e.g. a moth. [6], [15], [1]
Content-based Pricing based on the accessed content. [1]
QoS-based Pricing depends on the hired quality of service. [22], [1]
Location-based Pricing based on the access point of the user. [1]
Service type Pricing based on the usage of the service. [1]
Volume-based Pricing based on the volume of a metric (e.g. downloaded bytes). [22], [1]
Differentiation on
time-of-day
Pricing based on the hour when the service is used. [1]
Progressive Co-design Seller and buyer try to convene on a pricing plan. The seller announces a fixed price pair (p1, p2),
where p1 ¡ p2. Subsequently, the buyer commits a consumption level quality related to each price
announced and if agreed so he can buy additional units progressively if needed.
[6]
Competitor-Oriented (CO)
Pricing
At first, the vendor agent needs to select the competitor to compete with. Then, the vendor simply
decreases the price just below of the rival’s price. This algorithm requires perfect information of
the rival’s price.
[20], [22], [16]
Cost-based Following the approach of cost-based pricing, the price level is established using cost accounting.
According to it price determination based on costs can make good sense for SaaS.
[16]
Supply and Demand based In general way the unit price will vary until it settles at a point where the quantity demanded by
consumers (at current price) will equal the quantity supplied by providers (at current price),
resulting in an economic equilibrium of price and quantity.
[18], [23], [24],
[26], [19], [20]
Real-Time Pricing (RTP) Is a pricing model that dynamically changes its rate reacting to the classical supply and demand
rule, but with the difference that there is only one supplier. Amazon Web Services (AWS) offer a
simplified form of this pricing model called Spot Instances.
[26]
Derivative Follower Model It’s a kind of supply and demand based model simply adjusts prices by incrementally increasing
or decreasing them until the observed profitability level falls, then the direction of price
adjustment is reversed, thus seeking a local maximum of profitability.
[19], [20]
Hybrid Pricing Model This model allows a third entity called Price Authority dynamically adjust prices within static
limits to balance the workload on the basis of the queue wait times of jobs in grid environments.
[19]
Auction based Services are priced in an auction and usually carried out by a third party, called the market maker,
which collects the bids, selects the winners and computes the payments.
[17], [18], [1], [24]
English Auction All bidders are free to increase their bids exceeding other offers. When none of the bidders are
willing to raise the price anymore, the auction ends, and the highest bidder wins the item at the
price of his bid.
[24]
First-Price Sealed-Bid
Auction
Each bidder submits one bid without knowing the others’ bids. The highest bidder wins the item
at the price of his bid.
[24]
Vickrey Each bidder submits one bid without knowing the others’ bids. The highest bidder wins the item
at the price of the second highest bidder.
[24]
Dutch Auction The auctioneer starts with a high bid/price and continuously lowers the price until one of the
bidders takes the item at the current price. It is similar to a first-price sealed-bid auction because
in both cases the bid matters only if it is the highest, and no relevant information is revealed
during the auction process.
[24]
Double Auction In the double auction model, buy orders (bids) and sell orders (asks) may be submitted at any time
during the trading period. If at any time there are open bids and asks that match or are compatible
in terms of price and requirements (e.g., quantity of goods or shares), a trade is executed
immediately.
[24]
When performing the analysis, few studies explicitly stated
the formalization of SLA in Cloud/Grid Computing scenario.
However 4 primary studies ([11], [12], [14] and [21]) are
complementary. They are summarized following.
a) In [11] is introduced a framework that enables dynamic
specification and verification of SLAs on the Cloud. Its main
contribution to our research is an format of SLA-Description
based on XML specification which defines the main Quality
of Services (QoS) along with their threshold values agreed
up on selection of cloud services. It also defines the period
of service provision, the cost of using the service, and the
possible actions that should be taken if QoS provision is
frequently violated.
b) In [12] is presented a framework which the SLA pa-
6
Int'l Conf. Grid Computing and Applications | GCA'12 |
rameters are specified by metrics. These metrics define how
cloud service parameters can be measured and specify values
of measurable parameters. In addition to specific metrics this
study also propose general metrics that can be defined for SLA
with any or all types of cloud users.
c) In [14] the authors addressed the use of Cloud Comput-
ing for web hosting providers by creating a Cloud Hosting
Provider (CHP). They designed an SLA-aware web servers
management system in order to address the resources out-
sourcing mechanism on the provider’s part, defining important
economic variables to this kind of technology.
d) In [21] is proposed an unambiguous and flexible language
for formalizing SLAs and an architecture for specifying and
monitoring SLA’s on grid computing scenario. It references
a typical SLA formulated by Morris et al. [29] that includes
the components: Purpose,Parties,Validity Period,Scope,Re-
strictions,Service-Level Objectives,Service-Level Indicators,
Penalties,Optional Services,Exclusions and Administration.
V. A NALYSIS OF T HE RESULTS AND MAPPING OF STUDIES
By analysing the results, it can enable us to present the
number of studies tabulated in each category defined in this
study. Thus, it is possible to identify what have been empha-
sized in past research and determine gaps and opportunities
for future research [9].
A. Research Type Classification
TABLE V
RESEARCH TYPE CLASSIFICATION
Research Type Studies Quantity
Validation Research [14], [17], [18], [8], [22],[23], [25] 7 (30,4%)
Evaluation
Research [4], [20], [1], [24], [26], [25] 6 (26%)
Solution Proposal [11], [12], [13], [2], [15], [16], [3], [4],
[19], [21], [22], [5]
12
(52,1%)
Philosophical
Papers [6] 1 (4,3%)
Opinion Papers - 0 (0%)
Experience Papers - 0 (0%)
Initially, let us analyse the studies distribution regarding to
the research type classification (Table V).
It was notorious the “Opinion” and “Experience” papers
inexistence, while a number of “Validation”, “Evaluation” and
mainly “Solution Proposal” was found. Perhaps the rationale
was the contribution level desired by researches proposing
evaluable solutions to have more scientific relevance.
However, another more important point was observed re-
lated to “Evaluation Research”, it is notable the small quan-
tity of studies that matches this facet indicating insufficient
experimentation in industry.
Certainly there is progress in this direction, but the acceler-
ated growth in the cloud providers number (reported by [30])
influences the degree of competitiveness, causing the non-
disclosure of their proposals in the scientific community. This
fact encourages us to perform another research analysing cloud
provider’s accounting models in practice and comparing them.
B. Contribution Type Classification
Table VI shows the contribution type classification scheme,
which we can observe the most of studies propose concrete
“Models” or “Frameworks” instead of address activities re-
lated to accounting functions. This way, few “Processes”,
“Techniques” and none “Method” was registered. One possible
explanation may be the observation made earlier, regarding the
lack of practical results disclosed by the industry. In this case,
we can conclude that even small-scale, companies publish
“what they did” (models and frameworks) but hide the “how
they did” (processes, techniques and methods).
TABLE VI
CONTRIBUTION TYPE CLASSIFICATION
Contribution Type Studies Quantity
Method - 0 (0%)
Process [11], [4], [21], [5], [25] 5 (21,7%)
Technique [14] 1 (4,3%)
Model [13], [6], [2], [15], [16], [19],
[20], [1], [21], [8], [22], [23],
[24], [25]
14 (60,8%)
Framework [11], [12], [6], [17], [18], [3] 6 (26%)
C. Research Types X Research Questions
There were an effort in analysing the relationship between
the research questions and the research type, using a bubble
plot to represent the interconnected frequencies (Figure 1).
Fig. 1. Research type x Research questions
By analysing the chart upwards, none question was an-
swered by papers that addressed personal opinion or expe-
rience, on the other hand one paper gave a big contribution,
answering RQ2 and RQ3, classified as philosophical paper,
discussing a general pricing scheme that can be applied to
define variations for any computational element [12].
Most of the information (about 4 studies) related to account-
ing models comes from papers classified as “Solution Papers”.
Since this category includes 12 primary studies, it is clear that
there has been few research effort directed to the issue of
mitigating mechanisms aiming architecture improvements.
Related to RQ3, although the majority of studies was
classified as “Solution Proposal” and “Validation Research”,
Int'l Conf. Grid Computing and Applications | GCA'12 |
7
the papers doesn’t discusses how the pricing schemes could
be applied in a detailed way, unlike give just short concepts.
So, as our research aims to give a general overview mapping
the pricing schemes, future researches can focus on explain
how the pricing schemes can works in practice.
D. Accounting Models Analysis
The accounting models collected by this research were
categorized according to their features (see Table VII).
TABLE VII
ACCOUNTING MODELS ANALYSIS
Studies
Features
Pricing
Metering
Mediation
Accounting
Roaming
Billing
Charging
Finantial Clearing
Cloud Federation
Just in Time Clouds
User Interface
Security Support
SLA Support
Variable Payment Models
[1] XXXXXXXX XXX
[2] X X X X X
[3] X X
[4] X X
[25] X X X X X
Firstly, we used the taxonomy proposed by [5], aiming
to check what functions the proposed models used. Thus,
the terms pricing,accounting and billing appeared in more
than one paper and with the same meanings, which this
homogeneity indicates a certain taxonomy validity. Related to
accounting, two information stands out:
•In [1] the authors disambiguated the expressions account-
ing process and accounting function. Whereas accounting
process refer to a meta-concept that includes all the
taxonomy functions, accounting function is related to
recording and summarizing technical data in terms of
money, transactions and events;
•In [25] the accounting and billing functions are grouped
as integrated sub processes forming a type of macro-
process.
Lastly, it is important to highlight that the term billing was
cited by all primary studies. We attribute this result to the
influence of other areas such as telephony that has used largely
this term before cloud computing became a research trend.
Other features were derived from the most relevant aspects
found in primary studies. Cloud Federation was the first
feature. In this case it was observed a research gap in which
only one accounting model [2] were directed to federated
cloud infrastructures, needing to stress that this paper and [25]
belong to the same research group (the RESERVOIR project
[31]), showing as pioneer researchers in the area.
The feature Just in Time Clouds is a recent concept in which
providers only allocate resources when they are demanded
and only for the duration they are needed by their clients
[7]. To explore this mechanism showed promise, because
none accounting model addressed this feature. Something
previously expected, due to be a recent issue.
The User Interface Support was analysed, noticing that
some proposed models own a user interface that gives the
access control to managing accounting mechanisms on the
systems, but not all worried with this feature, only 40% of
them had a final user or admin user interface support.
In Security Support, just 60% of studies at least cited some
security mechanism like user authentication or transaction
authorization. When analysed SLA Support, it was verified if
the studies had SLA monitoring or the customer would choose
their service quality desired, noticing that, as such Security
Support, 60% fit this requirement. Therefore, SLA and Security
Support have been showed as relevant topics of interest in
accounting model field for cloud computing.
As last feature, it was investigated if the models were pre-
pared to support different payment models (Variable Payment
Models) such as Pre-Paid, Pos-Paid or Hybrid. These models
are in no way unique to clouds and on the contrary they are
well known to customers after being used for years in other
utility markets, most notably the mobile phone industry [25].
Hence some accounting models (40%) are ready, for exam-
ple, to work with resource consumption based on previous
purchased credits (Pre-Paid).
It has to be mentioned that initially it was thought to include
the term monitoring, however was preferred to use the term
SLA Support instead, due its less ambiguous concept. Accord-
ing to [25], SLA and monitoring are strictly related each other,
because the metric concept (from a monitoring point of view)
is very semantically close to the “Key Performance Indicators”
concept (from a SLA point of view).
Concluding, observing the fourteen features, one paper
had a greater coverage. The primary study [1] proposed a
flexible accounting model which can fit any service of cloud
computing that encompassed almost all features taken into
account by our classification. Therefore this paper can be used
as a starting point for future accounting models propositions.
VI. THREATS TO VALIDITY
There are some threats to the validity of our study, which
we briefly describe below.
•Research Questions: The research questions we defined
cannot provide complete coverage of the accounting field
related cloud and mainly grid computing, however, we
had several discussions to validate the questions.
•Publication Bias: We cannot guarantee that all relevant
studies were selected. We mitigated this threat as much as
possible, by following references in the relevant studies.
•Data Extraction: The studies were classified based on
our judgement, however, some studies could have been
classified incorrectly. To mitigate this threat, the classifi-
cation was performed by more than one researcher.
VII. CONCLUSION AND FUTURE WORK
We have introduced the results of a systematic mapping
study about accounting models for cloud computing investi-
gating scientific literature. In the end, starting from 580 papers,
23 filtered studies answered the research questions.
8
Int'l Conf. Grid Computing and Applications | GCA'12 |
As major contribution, this paper provides an overview
of the area and specific findings related to i) taxonomy for
accounting process, ii) accounting models, iii) pricing schemes
and iv) SLA composition.
i) The terms pricing,accounting and billing are the most
used terms. Among these, the term billing surely is the main
term in the area. This result is influenced by other fields such
as telephony that has used largely this word before cloud
computing became a research trend.
ii) In general there are few studies related to accounting
models for cloud computing, mainly in industry environment.
Besides there is a need for new proposals in federated cloud
infrastructures whereas the topics related to SLA and Security
have gained considerable attention.
iii) Despite the large amount of existing pricing scheme
types, there is a need in expose how they could be applied in
a detailed way, unlike give just short concepts.
iv) Related to SLA composition there are studies that
propose possible general items to compose the contract (e.g.
Scope,Penalties,Restrictions), others propose specific metrics
to monitor the services quality and others presents mechanisms
based on XML to specify metrics. Thus studying these results
it is possible to develop new solutions combining ideas.
Future work will focus on analyse more accurately these
mapping study results in order to match mainly the SLA
composition ideas with accounting processes/models found to
develop a more advanced accounting model. Also we intend to
study the use on real market of the pricing schemes identified.
ACKNOWLEDGEMENTS
This research was sponsored by the Program Center for the Re-
search and Development on Digital Technologies and Communication
(CTIC) of the Brazilian Minister of Science and Technology, grant
68/2012. In addition, this work was partially supported by the Na-
tional Institute of Science and Technology for Software Engineering
(INES1), funded by CNPq and FACEPE, grants 573964/2008-4 and
APQ-1037-1.03/08
REFERENCES
[1] I. R. Agundez et al., “A flexible accounting model for cloud comput-
ing,” in Proc. Global Conference (SRII 11). IEEE Computer Society,
Washington, DC, USA, Jul. 2011, pp. 277–284.
[2] E. Elmroth et al., “Accounting and billing for federated cloud infras-
tructures,” in Proc. Int. Conference on Grid and Cooperative Computing
(GCC 09), Aug. 2009, pp. 27–29.
[3] T. Pandey et al., “Authentication and billing framework for service
oriented architecture,” in Proc. Int. Conference on Systems, (ICONS 09),
Mar. 2009, pp. 91–95.
[4] K. W. Park et al., “Themis: Towards mutually verifiable billing transac-
tions in the cloud computing environment,” in Proc.Int. Conference on
Cloud Computing (CLOUD 10), Jul. 2010, pp. 139–147.
[5] I. Agundez et al., “A taxonomy of the future internet accounting
process,” in Int. Conference on Advanced Engineering Computing and
Applications in Sciences (ADVCOMP 10), Jun. 2010, pp. 111–117.
[6] A. Caracas and J. Altmann, “A pricing information service for grid
computing,” in Proc. Int. Conference on Middleware (MGC ’07), New
York, NY, USA, 2007, pp. 1–6.
[7] R. Costa et al. (2010) Just in time clouds: Enabling highly-elastic
public clouds over low scale amortized resources. [Online]. Available:
http://www.lsd.ufcg.edu.br/index.php/documentacao-menu#tr
1www.ines.org.br
[8] P. Samimi and A. Patel, “Review of pricing models for grid and cloud
computing,” in Proc. IEEE Symposium on Computers and Informatics
(ISCI 11), Mar. 2011, pp. 634–639.
[9] Petersen et al., “Systematic mapping studies in software engineering,”
12th Int. Conference on Evaluation and Assessment in Software Engi-
neering, vol. 17, no. 1, pp. 1–10, 2007.
[10] B. A. Kitchenham et al., “Cross versus within-company cost estimation
studies: A systematic review,” in Proc. IEEE Transactions on Software
Engineering, May 2007, pp. 316–329.
[11] A. A. Falasi and M. A. Serhani, “A framework for sla-based cloud
services verification and composition,” in Proc. Int. Conference on
Innovations in Information Technology (IIT), Apr. 2011, pp. 287–292.
[12] M. Alhamad et al., “Conceptual sla framework for cloud computing,” in
Proc. International Conference on Digital Ecosystems and Technologies
(DEST), Apr. 2010, pp. 606–610.
[13] M. Buthelezi et al., “Accounting, pricing and charging service models for
a guiset grid-based service provisioning environment,” in Proc. CSREA
EEE, 2008, pp. 350–355.
[14] J. O. Fito et al., “Sla-driven elastic cloud hosting provider,” in Proc. 18th
Euromicro Int. Conference on Parallel, Distributed and Network-Based
Processing (PDP 10), Feb. 2010, pp. 111–118.
[15] B. Jai, “The economy of parallel and distributed computing in the cloud,”
in Proc. Int. Symposium on VLSI Design, Automation and Test (VLSI-
DAT 2011), Apr. 2011, pp. 25–28.
[16] Lehmann et al., “Pricing strategies of software vendors,” Business and
Information Systems Engineering, pp. 452–462, 2009.
[17] M. Mihailescu and Y. M. Teo, “On economic and computational-efficient
resource pricing in large distributed systems,” in Proc. Int. Conference
on Cluster, Cloud and Grid Computing (CCGRID 10), Washington, DC,
USA, 2010, pp. 838–843.
[18] M. Mihailescu and Y. Teo, “Dynamic resource pricing on federated
clouds,” in Int. Conference on Cluster, Cloud and Grid Computing
(CCGrid), May 2010, pp. 513–517.
[19] R. M. Piro et al., “An economy-based accounting infrastructure for the
datagrid,” in Proc. Int. Workshop on Grid Computing(GRID 03), Nov.
2003, pp. 202– 204.
[20] Rohitratana et al., “Agent-based simulations of the software market
under different pricing schemes for software-as-a-service and perpetual
software,” in Economics of Grids, Clouds, Systems, and Services, ser.
Lecture Notes in Computer Science, Altmann et al., Eds. Springer
Berlin / Heidelberg, 2010, vol. 6296, pp. 62–77.
[21] A. Sahai et al., “Specifying and monitoring guarantees in commercial
grids through sla,” in Proc. Int. Symposium on Cluster Computing and
the Grid (CCGrid 2003), May 2003, pp. 292–299.
[22] J. Song et al., “Competitive pricing model for resource scheduling in
grid computing,” in Int. Conference on Semantics, Knowledge and Grid,
Oct. 2007, pp. 406–409.
[23] J. Yu et al., “A service-oriented accounting architecture on the grid,” in
Proc. Int. Conference Parallel and Distributed Computing: Applications
and Technologies (PDCAT 04), Dec. 2004, pp. 310–313.
[24] R. Buyya et al., “Economic models for resource management and
scheduling in grid computing,” in Concurrency and Computation Prac-
tice and Experience 14, Wiley Press, 2002, pp. 1507–1542.
[25] M. Lindner et al., “The cloud supply chain : A framework for infor-
mation, monitoring, accounting and billing,” in 2nd International ICST
Conference on Cloud Computing (CloudComp 2010), 2011.
[26] W. Sewook, “Debunking real-time pricing in cloud computing,” in Proc.
Int. Symposium on Cluster, Cloud and Grid Computing (CCGRID ’11),
May 2011, pp. 585–590.
[27] I. Freitas et al., “Agile software product lines: a systematic mapping
study,” Softw. Pract. Exper., vol. 41, pp. 899–920, July 2011.
[28] Wieringa et al., “Requirements engineering paper classification and
evaluation criteria: a proposal and a discussion,” Requir. Eng., vol. 11,
pp. 102–107, December 2005.
[29] R. Jain et al., “Congestion avoidance in computer networks with a
connectionless network layer,” Digital Equipment Corporation, MA,
Tech. Rep. DEC-TR-506, Aug. 1987.
[30] G. Jeremy. (2010, Oct.) The top 250 players in the cloud
computing ecosystem. [Online]. Available: http://cloudcomputing.sys-
con.com/node/1386896
[31] RESERVOIR. (2012) Resources and services virtualization without
barriers. [Online]. Available: http://62.149.240.97/
[32] M. Gohner et al. (2006) Accounting-ansatze im bereich des grid-
computing. [Online]. Available: http://tinyurl.com/87goucy
Int'l Conf. Grid Computing and Applications | GCA'12 |
9
