SERVICE LEVEL AGREEMENTS AS A SERVICE

Towards Security Risks Aware SLA Management

Katerina Stamou

, Jean-Henry Morin

, Benjamin Gateau

and Jocelyn Aubert

Institute of Services Science, University of Geneva - CUI, Geneva, Switzerland

Centre de Recherche Public Henri Tudor (CRPHT), Luxembourg City, Luxembourg

Keywords:

Cloud Computing, Service Level Agreement (SLA), Risk Management, Security, SLAaaS, Service Level

Objective (SLO).

Abstract:

Cloud computing has matured to become a valuable on demand alternative to traditional ownership models for

the provisioning of services, platforms and infrastructure. However, this raises many issues for Governance,

Risk and Compliance (GRC) and in particular in terms of Information Systems Security Risk Management

(ISSRM). Considering such issues lack attention and knowledge, particularly for small and medium sized en-

terprises (SMEs), and that cloud computing Service Level Agreements (SLA) provide very limited support

outside of basic Quality of Service (QoS) parameters, this paper argues that SLAs for cloud computing ser-

vices should be more customer oriented and aware of security and risk management. A design is proposed

where the SLA process, from context initialization to negotiation and agreement is decoupled from the actual

cloud service provisioning and itself turned into a Service : SLA as a Service (SLAaaS). This should provide

customers with much more customized and ﬁne-grained agreements compared with the ones currently offered.

1 INTRODUCTION

The cloud computing model allows easy, on-demand,

internet-based access to computing resources. NIST

(Mell and Grance, 2011) deﬁnes the following types

of cloud deployments: private, community, public, in-

cluding hybrid combinations of these. Each of them

involves different sharing modes (Jansen and Grance,

2011).

Provision of service offerings in cloud computing

is instant, requires only network access and minimal

client-provider interaction. Service level agreements

(SLAs) represent a form of such interaction, as guar-

antees between a service provider (SP) and a service

consumer.

SLAs provide assurances on the expected or mutu-

ally agreed service level on behalf of a SP to a service

consumer (Dan et al., 2003). They deﬁne a multi-

tude of business and technical Service Level Objec-

tives (SLOs), including metrics that reﬂect such ob-

jectives.

Still, current public cloud service offerings lack

the use of security risk management (RM), as well as

governance and compliance aspects in their provided

guarantees. We perceive RM as an important factor

to assist the ﬁne-tuning of an SLA according to cus-

tomer deﬁned objectives.

This paper advocates that SLAs for cloud com-

puting services should be more customer centric and

aware of customer risk requirements. We envisage a

holistic SLA process from context initialization to ne-

gotiation and agreement as a service that can provide

a customer with a much more tailored ﬁnal agreement

compared to the ones currently offered.

Section 2 presents the motivationbehind this work

and provides a high level summary of recent research

projects on cloud computing dealing with SLA man-

agement. We have identiﬁed some representative is-

sues regarding SLAs and their management. We also

brieﬂy discuss several derived SLA parameters and

management aspects.

Section 3 proposes a high-level SLA decomposi-

tion and clause classiﬁcation. This classiﬁcation has

helped conceptualize our approach and a model in a

research project on Cloud Computing SLA Risk Man-

agement (CLOVIS, for which a ﬁrst study was done

in (Morin et al., 2012)).

Section 4 discusses the importance of combining

RM with SLA management while decoupling it from

the actual provisioning of the service. We propose

a model and approach where SLA is provided as a

Service thereforeopening newopportunities for better

663

Stamou K., Morin J., Gateau B. and Aubert J..

SERVICE LEVEL AGREEMENTS AS A SERVICE - Towards Security Risks Aware SLA Management.

DOI: 10.5220/0003966006630669

In Proceedings of the 2nd International Conference on Cloud Computing and Services Science (CloudSecGov-2012), pages 663-669

ISBN: 978-989-8565-05-1

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

management of SLAs and consequently RM in cloud

services. We argue, such an approach may be partic-

ularly useful for SMEs having limited resources and

knowledge to assess such risks. We conclude with

on-going and future work.

2 SLAS IN CLOUD COMPUTING

2.1 Motivation

Cloud computing SLAs do not lack technical objec-

tives in terms of service levels. Undoubtedly, their

management lacks several technical attributes such as

machine-readability, automation etc. But most impor-

tantly, they lack risk management objectivesand more

generally GRC concerns that can be better addressed

with a more customer aware orientation.

Research approaches on SLA management intro-

duce important aspects like automation and dynamic-

ity, yet are deprived of RM objectives that are often

crucial to a customer’s internal planning. Although

commonly found SLA terms semantically relate to

GRC management, it is not clear how they relate to

SMEs’ needs and internal assessments.

Due to lack of experience, many SMEs and start-

up companies are not able to evaluate their risks and

often do not have a risk management plan. Typically,

their size and lack of ﬁnancial strength prohibits them

from doing so. Additionally, many SMEs primarily

look into public cloud service offerings, since they

cannot afford some other cloud deployment type.

Independently of the service layer, public clouds

provide SLAs that typically do not address small or-

ganizations’ security concerns. Thus, such agree-

ments impose even greater risks for SMEs that do

not have internal risk assessments. Moreover, pub-

lic cloud SLAs do not adjust their pricing according

to evaluation of speciﬁc risks, but apply massive, one-

size-ﬁts-all pricing models.

In (Potoczny-Jones, 2011) the offering of Secu-

rity Risk Agreements (SRAs) is proposed. Cloud SPs

offer such SRAs to SMEs customers as tailored SLAs

that include all necessary security information regard-

ing an offered service. An SME can then evaluate if

the offered service reﬂects its security requirements

or not. It is suggested that such agreements will in-

crease the level of trust between SPs and customers,

and consequently motivate them to adopt public cloud

services. Moreover, in (Kaliski et al., 2010) they pro-

pose the idea of Risk Assessment as a Service.

There is no doubt that the higher the mutual trust

between service provider and service consumer, the

easier the adoption of new offered services.

2.2 Related Work

We have conducted a thorough literature review to in-

vestigate state of the art approaches related to SLA

management for cloud computing services. Table 1

provides a high-level overview of recent research ini-

tiatives in cloud computing. Many initiatives are EU

FP7 funded projects and, as Table 1 illustrates, are fo-

cused on different cloud layers, with the exception of

(SLA@SOI, 2011) and (IRMOS, 2009) that follow a

more vertical approach across all layers.

Table 1: Research projects focused on different cloud com-

puting layers.

Project/cloud layer IaaS PaaS SaaS

Optimis x

Contrail x x

SLA@SOI x x x

IRMOS x x x

Cloud4SOA x

mOSAIC x

4CaaSt

Most studied SLA frameworks are platform spe-

ciﬁc, in that every research approach implements

its own system architecture and accommodates soft-

ware modules and tools. Such platforms typically in-

clude a multitude of software components to assist

ﬁne-grained interoperation of different services like

workﬂow (job) monitoring, resource discovery and

accounting. SLA management modules usually in-

clude active as well as template SLA repositories and

communicate with user and task management compo-

nents.

Automation of SLA management is a desirable

feature. Still, there is the assumption that at some

point human intervention is necessary. Depending on

the platform, layer and application speciﬁcs, some ap-

proaches implement a single type of SLA templates,

while others combine different types (IRMOS, 2009).

In CLOVIS, the goal is to deﬁne, design and de-

velop a service-level framework that integrates RM

and SLA capabilities. The focus is to improve gov-

ernance, risk management and compliance (GRC) as-

pects for cloud computing services while introducing

enhanced levels of managed ﬂexibility through excep-

tion management (Morin, 2008) features.

2.3 SLA Issues

Based on our review of related work, we have tried to

summarize important problems of currently provided

Morfeo 4CaaSt, http://4caast.morfeo-project.org

CLOSER2012-2ndInternationalConferenceonCloudComputingandServicesScience

664

SLAs. Undoubtedly, wide adoption of the cloud com-

puting model and increased interest about its offer-

ings by companies worldwide is leading to novel ser-

vice oriented economic models, that require more efﬁ-

cient, ﬂexible and automated SLA management (But-

ler et al., 2011).

Traditionally, SLAs are deﬁned exclusively by

SPs. The same SLA template is offered to all cus-

tomers of the same service, regardless of the speciﬁc

needs of customers. Such SLA approaches tend to ig-

nore that each customer weighs SLA deﬁned metrics

differently.

Offered SLAs focus on service level objectives

(SLOs) that deal with technical guarantees like up-

time, bandwidth availability, etc. Such SLOs are and

should be included in any service level agreement,

since they constitute the very basics of such an agree-

ment. On the other hand, such SLAs are monolithic

in the sense that no customer speciﬁc needs are taken

into account and their customization depends on man-

ual and time-consuming interventions between the

two involved parties.

In public clouds, SLAs do not address properly

important security aspects such as data privacy and

clearly lack negotiation opportunities between SPs

and customers. Provided SLAs are usually in non

machine-readable formats and are delivered as static

agreements, not considering dynamic demand on ser-

vice levels.

Moreover, it is commonly accepted that there is

a lack of uniﬁed standards for public cloud comput-

ing services (mOSAIC, 2011) and (Mell and Grance,

2009). The deﬁnition of standards can be very help-

ful for the establishment of common requirements and

provide an extra safety layer for SMEs.

Research efforts on SLA management for cloud

computing introduce important aspects like automa-

tion and dynamicity. They lack however risk man-

agement objectives that are crucial to a customer’s re-

quirements.

2.4 Important Aspects Regarding

Context and Management of SLAs

Our review of related cloud computing initiatives has

helped in aggregating important points considering

parameters and management of SLAs:

Machine-readable SLAs. Typically, most cloud of-

ferings for public cloud services offer static, non

machine-readable SLAs. All research approaches in-

dicate the necessity to offer machine-readable agree-

ments. This feature allows for automated negotiation

modules (SLA@SOI, 2011) as well as translation of

high-level QoS parameters into low-level ones (IR-

MOS, 2009).

Digitalization of SLAs construction and manage-

ment is a mature research topic. IBM’s research

on utility computing resulted in the deﬁnition of the

(WSLA, 2003) speciﬁcation language. The grid com-

puting community has also contributed much into

SLAs initiation and management for distributed in-

frastructures. The deﬁnition of the WS-Agreement

speciﬁcation language (Andrieux et al., 2005) as well

as the implementation of protocols related to SLA ne-

gotiation, resource allocation and monitoring (Cza-

jkowski et al., 2002) designate some grid-oriented

accomplishments that are currently reused by many

cloud computing activities.

Automation. Many research efforts highlightthe need

to automate as many SLA management parameters

as possible. Such automation is necessary to al-

low for vertical or multi-layered SLA management

(SLA@SOI, 2011). Moreover, it simpliﬁes the inter-

action of an SLA module with other system compo-

nents (monitoring, accounting etc) for the better or-

chestration of the whole SLA life cycle (Cloud4SOA,

2011).

Risk. (Optimis, 2011) refers to risk as a central con-

cept that has to be taken into account throughout the

full life cycle of the service. The Cloud Security Al-

liance provides a GRC Stack toolkit (CSA, 2010) that

integrates their initiatives for efﬁcient cloud auditing,

security assessments and critical compliance require-

ments.

Many SLA deﬁned terms and parameters are de-

rived, at least semantically, from RM techniques and

objectives.

SLA Management Activity Duration. Proposed SLA

frameworks and design approaches suggest that ac-

tivities related to SLA management should follow a

whole service life cycle. This feature enables dy-

namic management of SLAs by allowing their pos-

sible modiﬁcation during the run time of a service

(mOSAIC, 2011). Dynamic deviation of client de-

mands justiﬁes this requirement. Moreover, SLA ac-

tive duration affects other SLA management parame-

ters like pricing, a customer’s risk management plan-

ning, trust, etc.

Negotiation. Negotiation is a process that typically

takes place prior to the run time activity of the ser-

vice. (Optimis, 2011) highlights the need to negotiate

SLAs with different autonomous providers. (IRMOS,

2009) deﬁnes SLA renegotiation as negotiation dur-

ing the execution time of a service and describes it

as a process that takes place in order to solve issues

about to cause or having caused SLA violations.

SERVICELEVELAGREEMENTSASASERVICE-TowardsSecurityRisksAwareSLAManagement

665

Customization of SLA Related Terms and Parame-

ters. The (mOSAIC, 2011) project advocates the need

for user-centric, customer speciﬁc SLAs. Moreover,

(Contrail, 2011) categorizes SLA related terms into

quality of service and quality of protection parame-

ters. Such approaches indicate the need for differen-

tiated SLA offerings, depending on the speciﬁc needs

of a company, customized to express useful metrics.

Geographic Location, Inter/National Laws. The ge-

ographic location of data-centres in relation to the na-

tional and/or international laws is an important crite-

rion for many companies that want to invest into cloud

based solutions. It is commonly accepted that such

options should be included in, if not determine, an

SLA. As highlighted by NIST in (Mell and Grance,

2009) and other initiatives, currently there is a lack

of uniﬁed standards for cloud computing. The estab-

lishment of standards may assist in dealing with geo-

graphic and legally binding parameters.

Digital Signatures. In (IRMOS, 2009) they refer to

the eligibility of signing contracts using digital signa-

tures. They describe how they added signatures as an

extension to an agreement’s structure, which is based

on the WS-Agreement speciﬁcation (Andrieux et al.,

2005).

Regulation of Penalties. In (Contrail, 2011) they in-

vestigate a possible differentiation of penalties, based

on partial or complete fulﬁllment of SLA objectives.

They discuss penalties or rewards for violating or sat-

isfying an objective. Penalties can be imposed in

the form of monetary compensation that is the cur-

rent case in most public cloud service offerings or by

means of reputation track mechanisms (Rana et al.,

2008).

3 SLA DECOMPOSITION AND

CLAUSE CLASSIFICATION

In SLA Management Handbook Volume 2 (ITU,

2005), the authors deﬁne a method for classifying ser-

vice parameters to be used within an SLA. They clas-

sify parameters based on technology speciﬁc, service

speciﬁc and technology/service independent parame-

ters. Motivated by this methodology we have tried to

classify SLA clauses and context parameters by de-

composing an abstract SLA schema.

As core SLA clauses we have selected to use main

risk categories as identiﬁed in (ENISA, 2009). Such

a decomposition is very helpful to understand where

and how the RM factor can possibly be integrated

into an SLA. Additionally, it eases the process of rec-

ognizing particular needs and requirements that the

cloud computing model imposes. Moreover this de-

composition is quite practical for the realization of the

SLA model in CLOVIS.

Figure 1: SLA clause classiﬁcation.

Initially, our conceptual schema organizes SLA

clauses into technical, legal and policy organizational

elements. All such categories include sub-clauses that

are:

• derived from business opportunities

• service related or

• deﬁned by external requirements and standards

A subsequent layer follows where penalties are

deﬁned and regulated within an SLA.

This schema allows better mapping of companies

identiﬁed risks and easier service level updates –if

updates are allowed– during an agreement life cycle.

The holistic SLA process from construction to initial-

ization and management can be viewed as a modu-

lar procedure that introduces more customer oriented

characteristics.

The dynamic nature of the cloud computingmodel

substantiates the assumption that there is a tremen-

dous need for more customer oriented, risk man-

agement aware SLAs, combined with automated and

more ﬂexible characteristics. The proposed decom-

position illustrates a high level path towards the ﬁrst

part of such goals.

4 ISSRM COUPLED WITH SLA

MANAGEMENT AS A SERVICE

4.1 Proposed Design

A risk management tool coupled with an SLA man-

agement framework is likely to provide a valuable so-

lution for both customers and SPs. The SLA clause

decomposition and arrangement of Section 3 has in-

spired an abstract model that is illustrated in Figure

2. Both RM and SLA management are to be consid-

ered as provided ”as a Service”, hence the RMaaS and

SLAaaS.

CLOSER2012-2ndInternationalConferenceonCloudComputingandServicesScience

666

Figure 2: RM and SLA coupling.

Decoupling these components from the actual

cloud service provisioning as shown in Figure 3 has

many advantages. First and foremost it accounts for

a much needed separation of concern between ser-

vice provisioning and the conditions under which the

service is to be provided including monitoring. In

doing so, SLA management could be considered as

a Trusted Third Party (TTP) between SPs and cus-

tomers in a similar way Certiﬁcation Authorities are

now well recognized in similar trust roles. Moreover,

this may open a whole new area of expertise allowing

SLAaaS providers to become new marketplaces for

SLA management. One could even imagine indus-

try or service type speciﬁc SLA providers with tem-

plates tuned to speciﬁc needs. For example, a com-

pany working in the ﬁnance industry and requiring

a cloud storage service might be subject to speciﬁc

rules with respect to where the data physically resides.

In such a case, a Financial Services SLAaaS provider

could be a good choice for the company and addi-

tional peace of mind when contracting cloud based

storage services.

Figure 3: SLAaas decoupling from actual cloud service.

Our model assumes the existence of an appropri-

ate RM tool, which can provide a customer with an

adequate risk assessment according to a customer’s

service criteria. Theoretically, such criteria may de-

scribe service preferences from a single cloud layer

or from hybrid service combinations. The derived

assessment can subsequently be fed into a semi-

automated process that will map such information

into concrete SLOs.

It is not easy to automate a risk assessment pro-

cess, since every company and potential SLA cus-

tomer has unique and rather subjective requirements

regarding its risks and internal needs. Still, it is possi-

ble to use the output of a proper risk assessment pro-

cess and treat it as input for the orchestration of a cus-

tomer oriented SLA.

4.2 Service Usage Scenario

Let’s now consider the following basic service usage

scenario, described in abstract terms:

A customer (e.g., an SME), with or without an in-

ternal ISSRM plan, wants to invest into a suitable IaaS

solution. The customer is provided with a RM tool

that can help towards risk assessment.

The customer submits its risk assessment in raw

format along with its service preferences for the de-

sired service level guarantee (what type of service,

duration, IaaS common technical requirements). Our

SLAaaS framework receives as input the raw format

data and produces as output structured information

data in an interoperable format.

Such data represent the initial assessment, contex-

tualised into SLA clauses. These clauses are depicted

as SLOs. In (ITU, 2005) SLOs are deﬁned as internal

forms of SLAs that exist between business functions.

SLOs are less formal than SLAs.

Our output is not intended to produce a concrete

SLA, but rather context clauses that can directly be in-

tegrated in or affect a ﬁnal SLA output. It is merely an

attempt to assist the process of producing RM aware

and more customer oriented SLAs.

The processed output may then submitted to an

open registry that is accessible by SPs. SPs evaluate

the submitted objectives, adjust their own SLOs based

on their SLA templates and send their offers. Sent of-

fers are displayed back to the customer. The customer

may select or reject among the displayed offers.

4.3 Discussion

The Service usage scenario described above, assumes

several technical requirements. First of all, machine

readability of all involved information is an essen-

tial prerequisite. Information should be processed and

utilized in a well-used, interoperable format. A pro-

cessed SLO output should be able to ﬁt in most SLA

management frameworks, be editable and easily up-

dated.

In our current design, we are focused on the IaaS

layer. An initial part of our work will involve adapting

a selected RM tool into IaaS related risks. The report

provided in (ENISA, 2009) will support this effort.

Our hosting platform will include a web interface

for the customer (user) - service interaction as well as

a negotiation module for a possible customer-provider

interaction. Still, the envisioned service has also to

SERVICELEVELAGREEMENTSASASERVICE-TowardsSecurityRisksAwareSLAManagement

667

interact with several modules that are running on dif-

ferent platforms and assist services like monitoring,

accounting, etc.

Additionally, security is a major concern for our

framework since it has to process, move and regularly

update context sensitive, private data. Thus, our de-

sign will include the orchestration of the necessary

mechanisms to ensure data conﬁdentiality and secure

service transactions.

Moreover, a generalization of our envisaged ser-

vice equally applies to services of all cloud layers and

any hybrid combinations. We anticipate that the re-

alization of our model at the IaaS layer will generate

useful test cases and results that can be extended to

any cloud service approaches.

5 FUTURE WORK

Our tentative roadmap attempts to couple RM with

SLA management for offered cloud services. We plan

to validate our results using real case studies in collab-

oration with vendors who provide business resilience

solutions utilizing holistically the cloud computing

model. As mentioned, our produced case studies will

initially be applied to the IaaS layer.

The envisioned framework has to provide SLA

management capabilities as well as be compatible

with and pluggable into external platforms that may

integrate their own SLA management framework.

Compatibility will allow feeding of existing SLA

frameworks with RM parameters and metrics, so

that provided SLAs depict more precisely customers’

needs.

The produced SLA will be customer centric, tak-

ing into account and adjusting service level and pric-

ing parameters to a company’s identiﬁed risks and se-

curity concerns. This may lead to public cloud service

offerings that capture more thoroughly customers ob-

jectives while also being compliant with standards.

Currently the design process of our prospective

framework is initiated. We are in the process of se-

lecting and adjusting an appropriate RM tool to our

service scope. The next phase of our work will in-

clude the deﬁnition of a method that receives as in-

put a risk assessment output and maps it into potential

SLA clauses in a semi-automated manner.

Throughout this procedure we will use and reﬁne

the SLA decomposition schema of Section 3. Such an

arrangement can help with the desired SLA context

mapping as indicator of how to integrate and adapt

given risk assessment results. Additionally, we will

have to ensure that our method will allow for frequent

iterations and updates of the generated SLOs, since

the RM process is by essence dynamic.

The service we are building could possibly be

classiﬁed at the SaaS layer, since an end user would

ideally interact with the service through a web in-

terface. Still, inevitably the success and viability of

our proposed service depends on orchestrating com-

ponents and modules from lower layers of the cloud

computing stack.

6 CONCLUSIONS

We have provided an SLA decomposition and clause

classiﬁcation schema utilizing a method for service

parameter classiﬁcation that is provided in (ITU,

2005) and identiﬁed risk categories as identiﬁed by

(ENISA, 2009). Although abstract, such a classiﬁca-

tion may contribute as a ﬁrst step towards a uniﬁed

conception on SLAs construction, initiation and man-

agement for RM aware SLAs. Our assumption is that

more customer oriented, risk aware SLAs will assist

the integration of SMEs into the emerging cloud com-

puting economy.

Security concerns and often absence or negli-

gence of risk management operations are some of the

major obstacles that currently prohibit many SMEs

from engaging in and using cloud computing ser-

vices. Existing SLAs do not address such concerns.

Consequently, SMEs cannot estimate precisely their

needs regarding service level expectations and service

providers’ guarantees.

We have described our proposed design for the

SLAaaS framework. It will exhibit both risk and SLA

management attributes. A service-oriented frame-

work that addresses risk and generally GRC aspects

can lead to efﬁcient, more ﬂexible SLAs and be highly

valuable for many potential cloud service customers.

Finally, it can also be a very useful tool for service

providers to better adjust their pricing models and of-

fer negotiable, customer centric services.

ACKNOWLEDGEMENTS

This work is supported by the CLOVIS project jointly

funded by the Swiss SNF and Luxembourg FNR

Lead Agency agreement; under Swiss National Sci-

ence Foundation grant number 200021E-136316/1

and Luxembourg National Research Fund (FNR)

grant number INTER/SNSF/10/02. K.Stamou would

like to thank N.Mayer for external discussions and

valuable input in this effort.

CLOSER2012-2ndInternationalConferenceonCloudComputingandServicesScience

668

REFERENCES

Andrieux, A., Czajkowski, K., Dan, A., Keahey, K., Lud-

wig, H., Nakata, T., Pruyne, J., Rofrano, J., Tuecke,

S., and Xu, M. (2005). Web Services Agreement

Speciﬁcation (WS-Agreement). Available at http://

mailman.ogf.org/documents/GFD.107.pdf.

Butler, J., Lambea, J., Nolan, M., Theilmann, W., Torelli, F.,

Yahyapour, R., Chiasera, A., and Pistore, M. (2011).

SLAs Empowering Services in the Future Internet. In

The Future Internet, volume 6656 of Lecture Notes in

Computer Science, pages 327–338. Springer Berlin /

Heidelberg.

Cloud4SOA (June 2011). D1.3 Reference Architecture.

Technical report, EU FP7. Available at http://

www.cloud4soa.eu, accessed October 2011.

Contrail (2011). D3.2 SLA Management Services Terms

and Initial Architecture. Technical report, EU FP7.

Available at http://contrail-project.eu, accessed Octo-

ber 2011.

CSA (2010). GRC Stack, an Integrated Suite of

Four CSA Initiatives. Available at https://

cloudsecurityalliance.org/research/initiatives/grc-

stack, accessed November 2011.

Czajkowski, K., Foster, I., Kesselman, C., Sander, V., and

Tuecke, S. (2002). SNAP: A protocol for negoti-

ating service level agreements and coordinating re-

source management in distributed systems. volume

2537 of Job scheduling strategies for parallel process-

ing, pages 153–183. Springer.

Dan, A., Ludwig, H., and Paciﬁci, G. (2003). Web

service differentiation with service level agreements.

White Paper, IBM Corporation. Available at http://

www.ibm.com/developerworks/library/ws-slafram.

ENISA (2009). Cloud computing: Beneﬁts, Risks

and Recommendations for Information Se-

curity. Technical report. Available at http://

www.enisa.europa.eu/act/rm/ﬁles/deliverables/cloud-

computing-risk-assessment, accessed October 2011.

IRMOS (2009). D7.2.1 Initial version of Path Man-

ager Architecture and Guaranteeing QoS with Dy-

namic and Automated SLAs in real-time aware

SOIs. Technical report, EU FP7. Available at

http://www.irmosproject.eu, accessed October 2011.

ITU (2005). SLA Management Handbook, Concepts and

Principles, volume 2.0. TeleManagement Forum. Re-

lease 2.5, GB 917-2.

Jansen, W. and Grance, T. (2011). Guidelines on Secu-

rity and Privacy in Public Cloud Computing. Avail-

able at http://csrc.nist.gov/publications/nistpubs/800-

144/ SP800-144.pdf, accessed December 2011.

Kaliski, J., Burton, S., and Pauley, W. (2010). Toward Risk

Assessment as a Service in Cloud Environments. In

Proceedings of the 2nd USENIX conference on Hot

topics in cloud computing, HotCloud’10, pages 13–

13, Berkeley, CA, USA. USENIX Association.

Mell, P. and Grance, T. (2009). Effectively and Securely

Using the Cloud computing Paradigm.

Mell, P. and Grance, T. (2011). The NIST Deﬁnition of

Cloud Computing ( Draft ) Recommendations of the

National Institute of Standards and Technology. NIST

Special Publication, 145(6):7.

Morin, J.-H. (2008). Exception Based Enterprise Rights

Management: Towards a Paradigm Shift in Informa-

tion Security and Policy Management. volume 1 of In-

ternational Journal On Advances in Systems and Mea-

surements, pages 40–49.

Morin, J.-H., Aubert, J., and Gateau, B. (2012). Towards

Cloud computing SLA Risk Management: Issues and

Challenges.

mOSAIC (February 2011). D1.1 Architectural design of

mOSAIC’s API and platform. Technical report, EU

FP7. Available at http://www.mosaic-cloud.eu, ac-

cessed October 2011.

Optimis (2011). OPTIMIS SLA Framework and Term Lan-

guages for SLAs in Cloud Environment. Technical

report, EU FP7. Available at http://www.optimis-

project.eu, accessed October 2011.

Potoczny-Jones, I. (2011). Cloud Security Risk Agree-

ments for Small Businesses. Available at http://

corp.galois.com/blog/2011/8/23/cloud-security-risk-

agreements-for-small-businesses.html, accessed

November 2011.

Rana, O. F., Warnier, M., Quillinan, T. B., and Brazier, F.

M. T. (2008). Monitoring and Reputation Mechanisms

for Service Level Agreements. Grid Economics and

Business Models GECON, pages 125–139. Springer.

SLA@SOI (July 2011). D.A1a Reference Architecture for

an SLA Management Framework. Technical report,

EU FP7. Available at http://sla-at-soi.eu, accessed Oc-

tober 2011.

WSLA (2003). WSLA Language Speciﬁcation. Technical

report, IBM Corporation. Available at www.research.

ibm.com/wsla/WSLASpecV1-20030128.pdf, ac-

cessed October 2011.

SERVICELEVELAGREEMENTSASASERVICE-TowardsSecurityRisksAwareSLAManagement

669