Preliminary Design of a Platform-as-a-Service to
Provide Security in Cloud
Valentina Casola
1
, Alessandra De Benedictis
1
, Massimiliano Rak
2
and Umberto Villano
3
1
Universit
`
a “Federico II” di Napoli, Dipartimento di Ingegneria Elettrica e Tecnologie dell’Informazione, Napoli, Italy
2
Seconda Universit
`
a di Napoli, Dipartimento di Ingegneria dell’Informazione, Aversa, Italy
3
Universit
`
a del Sannio, Dipartimento di Ingegneria, Benevento, Italy
Keywords:
Cloud, Cloud Security, SLA, Requirements.
Abstract:
Cloud computing is an emerging paradigm, widely adopted in distributed and business computing. Never-
theless, the biggest issue with the large adoption of cloud computing is the perception of loss of security and
control over resources that are dynamically acquired in the cloud and that reside on remote providers, and the
strong integration of security mechanisms into system architectures. This paper deals with the integration of
security features into cloud applications by an as-a-service approach, using Service Level Agreements as a
means to clearly define rights and constraints of both customers and providers. The goal is to show the main
requirements of a platform dedicated to security and to present the global architecture, in terms of components
and their interactions, devoted to negotiate, monitor and enforce the security mechanisms to be applied over
existing cloud providers.
1 INTRODUCTION
Cloud computing, based on features as on-demand
self-service and on the pay-per-use business model, is
currently the emerging paradigm for distributed com-
putation and has been recently indicated as a strate-
gic asset by the European Commission (Commission,
2011), which is assuming a set of initiatives (e.g. Eu-
ropean Cloud Partnership (Commission, )) aimed at
supporting its adoption in both private and public bod-
ies.
The biggest issue with the large adoption of cloud
computing is related to security (Dekker, 2012; Cat-
teddu, 2011), as the paradigm assumes that all re-
sources are delegated to the cloud (i.e., hosted and
managed by a remote cloud provider, and can be
hardly ever physically located), while typically cloud
providers have their own policies for the management
of reserved data and privacy. The approach almost
universally followed to define guarantees on the pro-
visioning of a service is the introduction of Service
level Agreements (SLAs). SLAs should specify Ser-
vice Level Objectives (SLOs) dedicated to security,
but there is still a lack of standards for the definition
of the security terms in a SLA, even if a lot of work
is ongoing in this field by dedicated standard groups
(as the SLA C-SIG from the European Commission,
the CSCC SLA group(CSCC, )) and research projects
(see CUMULUS (Pannetrat et al., 2013), A4Cloud
(Pearson, 2011), and SPECS (Rak et al., 2013)).
In addition to the above discussed issues, a fur-
ther problem is linked to the “traditional” design of
security systems, which often requires full control of
the systems to be protected, in contrast with the cloud
principles that would require an approach based on
modular security (Battista et al., 2013). To this end,
the FP7 has funded the SPECS Project (Rak et al.,
2013), which proposes a cloud Platform-as-a-Service
(PaaS) dedicated to provide Security-as-a-Service us-
ing SLAs. In this paper, we present a preliminary
requirement analysis for platforms like SPECS and
identify the main involved components. In order to
clarify the whole analysis process, we illustrate the
simple methodology adopted to derive requirements,
which can be successfully applied to the design of any
PaaS.
The remainder of this paper is organized as fol-
lows. Section 2 illustrates the context of our work,
while in Section 3, we present an overview of most
relevant PaaS solutions and of the proposed platform.
In Section 4 we show, for the most representative
functionalities we are interested at offering, the whole
extraction process of requirements. Finally, in Section
5 a preliminary platform architecture is presented, and
752
Casola V., De Benedictis A., Rak M. and Villano U..
Preliminary Design of a Platform-as-a-Service to Provide Security in Cloud.
DOI: 10.5220/0004979307520757
In Proceedings of the 4th International Conference on Cloud Computing and Services Science (MultiCloud-2014), pages 752-757
ISBN: 978-989-758-019-2
Copyright
c
2014 SCITEPRESS (Science and Technology Publications, Lda.)
in Section 6 we draw our conclusions and identify fu-
ture work directions.
2 MOTIVATION AND CONTEXT
In this Section, we provide some representative sce-
narios that outline the main open issues related to the
management of security requirements in a cloud en-
vironment, and help to evidence the main functional-
ities required from a security-oriented platform. As a
first scenario, let us consider a cloud federation whose
set-up is explicitly subject to security requirements
(i.e. the federated providers must ensure a set of se-
curity features). At the state of the art, the evaluation
of security attributes related to a cloud provider is car-
ried out by hand (for example, based on the provider’s
existing ISO/IEC 27000-series certifications), as there
are no automatic tools to search, among the avail-
able cloud offerings, those matching specific security
attributes. This is mostly due to the lack of stan-
dards to specify the requested/provided security pa-
rameters. Recently, some steps have been taken for
improving transparency and assurance in the cloud
by the Cloud Security Alliance, with its STAR (Secu-
rity, Trust & Assurance Registry) initiative. However,
even if STAR is a powerful means to help users assess
the security of cloud providers, there is still a lack of
mechanisms to effectively evaluate and compare dif-
ferent offerings, given a specific request.
Similar issues can show up even in the traditional
interactions between cloud providers and end-users.
Let us consider, for example, a cloud end-user that
wishes to acquire a secure cloud storage service to
remotely store data with specific data confidential-
ity requirements. With the currently available ap-
proaches, the user has first to manually search for
cloud providers with the desired storage features, then
check their SLAs in order to verify the kind of con-
fidentiality features offered, study the different avail-
able offerings, and finally select the one that fulfills as
much as possible its requirements. As for the previous
case, it would be desirable to provide the end-users
with a single access point to specify their functional
and non-functional (security) requirements and with
automatic tools to search for and to select the best of-
fering. Moreover, in both scenarios, once a security
control has been guaranteed from a provider, the cus-
tomer should be provided with proper tools to monitor
the correct service provisioning while, at the state of
art, the customer must accept the provider monitoring
services and metrics, and can only create local mea-
surements to obtain further information (e.g., custom
local measurements of response time).
3 DEFINING A
SECURITY-ORIENTED PAAS
As stated, our main goal is the design and the imple-
mentation of a PaaS dedicated to security services,
based on an SLA approach. At the state of the art,
PaaS solutions are very different from one another,
and there is no standard interpretation of how to build
them. The most common solution is based on adapt-
ing application servers (like Apache Tomcat), in order
to decouple application deploying from virtual ma-
chines acquired over Infrastructure-as-a-Service so-
lutions. Examples of such solutions are offered by
FP7 projects as cloud4SOA (Zeginis et al., 2013),
OPTIMIS (OPTIMIS, ) or Contrail (in the ConPaaS
module (Pierre and Stratan, 2012)). Such PaaS solu-
tions are usually dedicated to a single technology (in
the most common case, web applications), which is
cloud-enabled in a way completely transparent to ap-
plication developers. This approach is based on the
idea of porting well-known middlewares to the cloud,
and of making transparent to middleware users the
adoption of the cloud paradigm for lower level re-
source acquisition. In other words, applications (and
their developers) are not aware of the cloud.
An alternative approach is the one proposed by
integrated stacks as Google App Engine (GAE) or
Microsoft Azure, where developers must adopt dedi-
cated APIs to build cloud-enabled applications. Such
platforms are integrated with the offering of the CSP,
who manages the full stack of services (SaaS, PaaS
and IaaS). Applications are developed using dedi-
cated toolkits: in this case, application developers are
aware of the adoption of the cloud paradigm and have
a set of dedicated APIs that enable them to exploit
the cloud flexibility in their applications. As a draw-
back, applications are dependent on the platform, and
cannot be used over different cloud providers without
modifications.
Finally, a further approach has been proposed in
solutions as mOSAIC (Petcu et al., 2013) or open-
shift (RedHat, ). These adopt a dedicated cloudware,
which can be deployed over infrastructure resources
and offers APIs similar to the ones offered by GAE
or Azure, but independent of the CSP that delivers the
(infrastructure-level) resources. Our goal is to start
from such solutions and build a PaaS dedicated to
offering security (Security-as-a-Service) through an
SLA-based approach.
As the first step towards the design of the plat-
form, we analyzed the possible interactions with
prospective customers, identifying three interaction
models illustrated in details in Section 3.1. After
that, in order to define the platform requirements, we
PreliminaryDesignofaPlatform-as-a-ServicetoProvideSecurityinCloud
753
adopted a simple methodology based on the following
steps:
we extracted the required macro-functionalities
from some example scenarios described in Sec-
tion 2 and, for each macro-functionality, we iden-
tified the involved actors and roles, and the differ-
ent usage perspective deriving from the different
interaction models defined in the previous phase;
we described each usage perspective by means of
an high-level interaction diagram, pointing out the
main actions performed by the platform; from this
functionalities we identified a preliminary set of
high-level use cases from the previous interaction
diagrams;
we classified the resulting use cases into different
service layers and carried out a detailed analysis
of each service to identify lower level use cases.
Starting from the above described requirement
analysis, we identified a preliminary architecture that
subdivides the services offered by the platform in dif-
ferent modules and outlines their interactions. Section
5 will describe such preliminary design.
3.1 Actors, Roles and Interaction
Models
The platform is intended for different users and can
serve different purposes: both end-users and CSPs
can access it in order to (i) specify their cloud secu-
rity requirements by means of SLAs, (ii) automati-
cally evaluate and compare security features offered
by remote CSPs (brokering of security services), (iii)
enhance a remote service based on specific security
requirements, (iv) manage the full SLA life-cycle (ne-
gotiation, monitoring and enforcement), and (v) de-
velop and deploy security services that are cloud SLA-
aware, implemented as a flexible open-source PaaS.
Such functionalities imply that the involved ac-
tors can play different roles depending on the specific
functionality they invoke (e.g., developer, adminis-
trator, customer . . . ). Moreover, based on the spe-
cific hosting configuration and on the resulting dif-
ferent security and responsibility issues, we can iden-
tify three different ways to interact with the platform,
which introduce further role classifications. In or-
der to have a clear and shared role definition, we
will adopt the high-level stakeholder taxonomy pro-
posed in the Cloud Standards Coordination (CSC) ini-
tiative launched by ETSI (ETSI, 2013), which ba-
sically considers the roles of customers, providers,
and partners for each party (represented by either an
individual or an organization). While customers and
providers are those asking for a service and offering
SPECS
End User
Services
Improving
Security
Services
Secure
Brokering
Services
Cloud
Provider
used by Customers
secured by SPECS
used by Customers
chosen by SPECS
used by SPECS
Cloud
Provider
Cloud
Provider
SPECS
Developer
SPECS
Administrator
Customers
Figure 1: Interaction Model 1: Third Party Security Plat-
form.
the service, respectively, partners are those providing
support to the provisioning and/or the consumption
of services. To avoid ambiguity, in our discussion
we will refer to the end-users as the parties located
(as customers) at the beginning of the service sup-
ply chain, while the external providers will represent
those CSPs located at the end of the supply chain. We
do not care about the way these CSPs offer their ser-
vices, which are in the following considered atomic.
Even if this approach is very general, and cov-
ers all the different aspects of cloud service supply
chains, it has the side effect that in different sce-
narios the same actor can play different roles, based
on the considered perspective, generating confusion
in the requirement analysis of a single cloud product
that can be deployed in different ways. In particular,
the platform can be offered in three different ways as
shown in the following interaction models.
The main interaction model, named Third Party
Security Platform (IM1), is proposed in Figure 1.
The platform runs as an independent third-party com-
ponent, consuming its own resources – acquired from
a public or private cloud provider and managed by the
administrator – and offering its services to end users.
In such model, the platform needs to authenticate its
own users and to manage their credentials in a secure
way. As for responsibility taking, it depends on the
role played by the platform. Indeed, the platform can
play both the ETSI roles of CSP and partner: as a
CSP, it signs a contract with the end user and takes
care of the entire remote service invocation process
on his behalf. When it acts as a partner, instead, the
user only exploits its brokering capabilities to find the
service that bests suits his needs, but then signs a con-
tract with the selected remote service provider to have
his requirements granted.
The second interaction model is named Hosted
CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience
754
Other Services
SPECS
End User
Services
Secure
Brokering
Services
Local
Resources
SPECS & Local Resource
Administrator
used for (secure)
Cloud Bursting
Cloud
Provider
Improving
Security
Services
Customers
Figure 2: Interaction Model 2: Hosted Platform.
Platform (IM2) and is shown in Figure 2. In such
interaction scheme, the administrators are co-located
within the Hosting CSP, which internally hosts the
platform. A hosting CSP can exploit the platform in
order to (i) enhance security features of the services
and resources being normally offered to end users, (ii)
acquire external CSP services to fulfill users’ requests
(in this case, secure brokering services can be used
also to perform cloud bursting note that such ser-
vices are invisible to end users), and offer additional
functionalities to users by re-selling the end-uses ser-
vices implemented by the platform. Differently from
the first interaction model, in this case the platform
acts only as a CSP, and never as a partner, because it
directly provides the hosting CSP with the requested
services taking on all responsibilities for them.
The third interaction model is referred to as User
Software (IM3) and is shown in Figure 3. It assumes
that the platform is dedicated to a single end user, who
installs and runs it to manage his own activities. In
this case, the platform does not offer services to other
end users, but it is just adopted by its administrator
(the end user who owns it) in order to manage secu-
rity features over remote services. For this reason,
internal services are almost useless (no need for au-
thentication, or for offering something internally ded-
icated), while brokering and security enhancement are
the only exploited functionalities. In this interaction
model, the platform acts only as a partner, because
the end user adopts its features only to build his own
services.
4 PRELIMINARY
REQUIREMENT ANALYSIS
In this Section we show the whole process followed
to analyze some the main functionalities offered by
the platform to derive a preliminary set of require-
ments. In particular, for the sake of brevity, we fo-
SPECS
End User
Services
Secure
Brokering
Services
Improving
Security
Services
End User and
SPECS Administrator
used by Customers
secured by SPECS
Cloud
Provider
Cloud
Provider
used by SPECS
Figure 3: Interaction Model 3: User Software.
Table 1: Brokering usage perspectives.
Interaction Models Brokering
IM1 (as Partner)IM1-P / (as
CSP)IM1-CSP
IM2 (as CSP) IM2-CSP
IM3 (as Partner)IM3-P
cus on one of the macro-functionalities identified in
Section 3, that is Brokering of an IaaS (Service) with
granted security parameters.
With respect to this macro-functionality, we ana-
lyzed the behavior of the platform in the three inter-
action models proposed in the project. In particular,
for each interaction model, we considered the possi-
ble roles played by the platform (CSP or Partner-P),
deriving a set of usage perspectives, shown in Table
1. As illustrated in Section 3.1, not all the combi-
nations are allowed: for example, in the interaction
model 2, the platform acts only as a CSP and never
as a Partner, because it directly provides the Hosting
CSP with the requested services taking on all respon-
sibilities for them. In the following, we are going to
detail a couple of usage perspectives, and report high
level interaction diagrams from which to derive a pre-
liminary set of use cases.
In Figure 4, we report the high level interaction di-
agram for the usage perspective named IM1-CSP: the
aim is to identify the basic steps related to the execu-
Figure 4: The platform acts as a broker in IM1.
PreliminaryDesignofaPlatform-as-a-ServicetoProvideSecurityinCloud
755
tion of the brokering macro-functionality in the case
when the platform takes on all responsibilities on the
quality (i.e. level of assured security) of the invoked
services. As depicted, the involved actors are the Cus-
tomer, the Platform-CSP and one or more External
CSPs. We assume that the Platform-CSP manages a
repository of existing CSPs and related SLAs and a
customer submits to the Platform-CSP a SLA request
with a list of security requirements. The basic flow of
events can be described as follows:
1. a customer submits to the Platform CSP his secu-
rity requirements in terms of a SLA,
2. the Platform CSP compares the SLAs security re-
quirements against those specified in its CSP SLA
repository and possibly applies a ranking algo-
rithm,
3. the Platform CSP negotiates with the customer the
target CSP to invoke based on its SLA,
4. the Platform CSP and the customer sign the SLA,
and the Platform CSP stores the signed SLA in a
signed SLA repository,
5. the customer invokes one of the negotiated ser-
vices on the Platform CSP, which forwards the in-
vocation request to the proper access point and re-
trieves possible results, to return them back to the
customer.
With respect to the other interaction models, the bro-
kering function is basically carried out in a similar
way, except that no SLA is signed between the cus-
tomer and the platform whenever the latter acts as a
Partner.
From the analysis of this scenario, we were able to
derive a first set of functionalities, which can be log-
ically split into three service layers: the application
service layer includes all services related to the im-
plementation of the business logic of the application
built on top of the platform (in this case, the appli-
cation that offers a broker of secure services). The
core service layer includes all services related to ne-
gotiation of SLAs and search, evaluation and ranking
of providers based on their SLAs. Finally, the Plat-
form service layer includes all basic enabling services
(e.g.set-up, activation, deployment of services on the
cloud infrastructure) and services related to SLA life-
cycle management.
5 PRELIMINARY
ARCHITECTURE DESIGN
A preliminary architecture design is shown in Fig-
ure 5: as illustrated in the previous Section, the ser-
vices offered through the platform can be classified
into three service layers: the Application layer, the
Core Services layer and the Platform layer. The ap-
plications built on top of the platform belong to the
application layer and are implemented by using the
services offered by both Core and Platform layers.
The main concept behind the platform is that every
functionality is offered as a service and all applica-
tions are managed through SLAs. For this reason,
the interactions among an application hosted by the
platform and its customers are subject to negotiation
and controlled via enforcement and monitoring func-
tionalities. According to this view, the Core Services
layer is composed by the following modules:
Negotiation Module: manages all the services
dedicate to guide customers and providers to
reach the agreement on SLAs.
Enforcement Module: manages all the services
dedicated to apply security mechanisms and cor-
rectly configure and use services, on the basis of
the agreed SLAs.
Monitoring Module: manages all the services
dedicated to collect and evaluate the state of the
services acquired.
At the Platform layer, as discussed in the pre-
vious Section, we locate both low-level (enabling)
services, to support the correct execution of higher-
level services in the cloud environment and offer basic
functionalities (repository management, SLA man-
agement), and services for interoperability, which im-
plement the actual communication among services lo-
cated at the core layer. Indeed, architecture modules
must be usable independently one of the others, pro-
vided interfaces among modules will be as simple as
possible and all modules will support the SLA-based
philosophy. This implies that the communication
among services of the Core layer will be possible only
through the Platform layer services, which will also
support the interoperability among modules. From
the architecture point of view, the Platform layer will
be then composed of two sub-layers:
Enabling Platform: enables the execution of com-
ponents (configuration, starting, stopping, pack-
aging) on-top of a cloud-like (IaaS-like) infras-
tructure and offers other generic support ser-
vices(component configuration, communication,
storage).
SLA Platform: offers services to enable the man-
agement of SLA life cycle, including the man-
agement of SLA repositories and the communica-
tion among services dedicated to different phases
of SLA life cycle, acting as the SLA-based glue
CLOSER2014-4thInternationalConferenceonCloudComputingandServicesScience
756
Figure 5: Architecture.
for the security services we are interested to offer
over the platform.
The layers also put in evidence that SPECS is
able to provide both Software-as-a-Service (SaaS)
and PaaS service models: as a SaaS, it offers a de-
ployed Platform, hosting a set of applications, which
provides application-specific services to its own cus-
tomers; as a PaaS, it offers a deployed Platform that
enables the deployment of new applications. As an
example, from one side the customers can use the
platform to invoke (use) an Improved Security Ser-
vice, consisting in a service either locally developed
in the platform or obtained by improving an existing
service offered by a remote CSP, which has specific
security features. From the other side, the customers
can use the platform to develop, for example, an Im-
proved Security Service, based on basic enabling ser-
vices (those offered by the Enabling Platform and the
SLA Platform).
6 CONCLUSIONS
In this paper we presented the preliminary analysis
of a cloudware dedicated to create a PaaS that of-
fers security services through an SLA approach. We
proposed a methodology to design such kind of plat-
forms, and sketched the preliminary results of the re-
quirement analysis, starting from some of the usage
scenarios identified in the context of the SPECS FP7
project. We argue that the proposed methodology is
of general interest for the design of cloudwares and
dedicated PaaS security solutions. The requirement
analysis presented here can be reused independently
of the technology which will can be involved, offering
a high-level view over the provision of security as-a-
service using SLAs. Moreover, we proposed a pre-
liminary design of the Platform, based on the consid-
erations and the classification of services made in our
preliminary analysis. We plan to upgrade this require-
ment analysis in the future, collecting feedbacks from
cloud stakeholders to refine the requirements and the
preliminary design, and to start the prototyping phase
of the platform.
REFERENCES
Battista, E., Casola, V., Mazzocca, N., Ficco, M., and Rak,
M. (2013). Developing secure cloud applications:a
case study.
Catteddu, D. (2011). Security & resilience in governmental
clouds. Technical report.
Commission, E. European cloud partnership.
http://ec.europa.eu/digital-agenda/en/european-
cloud-partnership.
Commission, E. (2011). Unleashing the potential of cloud
computing in europe. Technical report.
CSCC. The cscc practical guide to cloud service level
agreements. Technical report.
Dekker, M. (2012). Critical cloud computing a ciip perspec-
tive on cloud computing services. Technical report.
ETSI (2013). Cloud standards coordination. Technical re-
port.
OPTIMIS. The optimis project web site.
http://www.optimis-project.eu/.
Pannetrat, A., Hogben, G., Katopodis, S., Spanoudakis, G.,
and Cazorla, C. (2013). D2.1: Security-aware sla
specification language and cloud security dependency
model. technical report, certification infrastructure for
multi-layer cloud services (cumulus).
Pearson, S. (2011). Toward accountability in the cloud. In-
ternet Computing, IEEE, 15(4):64–69.
Petcu, D., Martino, B. D., S., V., Rak, M., Mhr, T., Lopez,
G. E., Brito, F., Cossu, R., Stopar, M., Sperka, S.,
and Stankovski, V. (2013). Experiences in building
a mosaic of clouds. JOURNAL OF CLOUD COM-
PUTING, 2:–.
Pierre, G. and Stratan, C. (2012). Conpaas: a platform
for hosting elastic cloud applications. IEEE Internet
Computing, 16(5):88–92.
Rak, M., Suri, N., Luna, J., Petcu, D., Casola, V., and Vil-
lano, U. (2013). Security as a service using an sla-
based approach via specs. In IEEE, editor, Proceed-
ings of IEEE CloudCom Conference 2013.
RedHat. Openshift web site. https://www.openshift.com/.
Zeginis, D., D’Andria, F., Bocconi, S., Cruz, J. G., Martin,
O. C., Gouvas, P., Ledakis, G., and Tarabanis, K. A.
(2013). A user-centric multi-paas application manage-
ment solution for hybrid multi-cloud scenarios. Scal-
able Computing: Practice and Experience, 14(1).
PreliminaryDesignofaPlatform-as-a-ServicetoProvideSecurityinCloud
757