Security Criteria in Deciding on Migration of Systems to
the Cloud
Rafael Gómez
1
, David G. Rosado
2
, Daniel Mellado
1
and Eduardo Fernández-Medina
2
1
Spanish Tax Agency, Madrid, Spain
2
GSyA Research Group, University of Castilla-La Mancha,
Dept. of Information Systems and Technologies, Ciudad Real, Spain
Abstract. Cloud computing is setting trend in IT world. As it evolves,
providers and clients claim their concern about their pros and cons. Some
proposals have been made on the methodologies to assess criteria for benefits
and risks of the different cloud models. How these proposals deal with security
issues (that most IT executives point out as their top concern)? In this paper we
go into the issue of how we can incorporate security requirements to a decision
making process for whether to migrate legacy systems to the cloud and how to
do it. From systems in control of the firms’ data centers to systems working
partially, if not totally out of their control.
1 Introduction
Mell & Grance propose in [20] the most widespread definition for the ultimate utility
model in the IT field: the cloud computing. They describe cloud computing as “a
model for enabling ubiquitous, convenient, on-demand network access to a shared
pool of configurable computing resources that can be rapidly provisioned and
released with minimal management effort or service provider interaction.”
Some like Jansen & Grance say that cloud might be “at odds with traditional
security models and controls” [15] and, for IT executives, security is a top concern
when they are asked about the major deterrents for adoption of the cloud model [8,
10]. So it's still today [23] (see fig. 1).
But, is this concern really undermining the final decision when we face in the
other plate of the balance the promise of huge savings? How can we model and
calibrate the pros and cons of the whole migrating process of a business? After all,
security criteria are probably some of the most difficult to size of all criteria in the
decision process.
There are authors who see the cloud not as a threat but an opportunity to improve
the security of many legacy systems. Winkler says in [25] that the migration to the
cloud “gives us hope that we can regain control over gaps and issues that stem from
poorly integrated or after-thought security.” Also, Buyya et al. point out in [4] that
Gómez R., G. Rosado D., Mellado D. and Fernández-Medina E..
Security Criteria in Deciding on Migration of Systems to the Cloud.
DOI: 10.5220/0004099600930100
In Proceedings of the 9th International Workshop on Security in Information Systems (WOSIS-2012), pages 93-100
ISBN: 978-989-8565-15-0
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
security can be guaranteed more easily over grids and clusters.
According to a research carried out by MeriTalk among 166 US Federal IT
executives 47% of the existing federal IT applications are based on legacy
technologies that need modernization [24]. The overall federal budget to maintain
this kind of systems can be estimated around the $35.7B, somewhat around 40% of
the overall federal budget for IT (this percentage when applied to the whole world IT
annual budget stands out up to $1.45T). According to Gartner, modernization
technologies for legacy is the fourth in the list of IT executives' priorities [11].
In this paper we present a preliminary research of some of the efforts that have
been made in the field and the concrete references that we have found regarding
issues like the impact of the security criteria in the migration of legacy systems to the
cloud. It is meant to be only the first step into a full research aiming the problem and
how it is presented to private and public organizations, what issues they regard as
important, and how they incorporate security criteria on their decisions.
The paper is structured in seven sections. The first one is this introduction. Section
two presents a definition on the decision problem. Sections three and four present two
related issues: the current state on multi-criteria analysis and the current research on
the legacy systems migration. In section six gives out some related works. Last, in
section seven we present our conclusions and future work.
Fig. 1. Top concerns surrounding cloud computing.
(source The Open Group Cloud Computing Survey 2011)
2 Problem Definition
An organization that plans migrating its legacy systems to the cloud must face several
challenges: to reduce or keep the overall provision costs while ensuring the functional
levels and enhancing or maintaining some other requirements like scalability, security
and privacy, and response when a lack of availability event comes about.
Decision Makers: Most of the times, the leading decision makers are the non IT
executives (i.e. those in the functional areas) and also those who are responsible of the
IT area. Usually all of them work together since there are common interests though
each one might have different concerns about the migration goals.
Goals: In any case, the goal of the decision is not unique and, frequently, they are
conflicting goals. Every project of migration of a system to the cloud (whether or not
is legacy) aims one or more of these:
16
20
17
26
19
2
Cos t
Interoperability & portability
Performance issues
Security
Vendor lock-in
Other
94
To reduce the cost associated to the normal work of the system.
To improve the flexibility of the system to cope with peak times.
To improve the system availability.
To ease the sustainable growth of the system.
To improve the intrinsic security of the system.
To prepare the organization for a (may be partial) outsourcing of IT function.
Uncertainty and Risks: On the other hand, we must be prepared to accept that either
decision (migrate it or not, and which systems are to migrate, and which model we
choose) have some uncertainties that we must manage. For instance, the main
uncertainty usually comes from the financial side. A sustained growth of the
organization might justify investments and service costs of traditional data centres,
but if a recession strikes, maintaining oversized systems may impose an excessive
cost impossible to reduce with traditional models. Another uncertainty we find is the
question on how technology evolves and the lower costs that the evolution brings.
Impact of Time: A legacy migration (valuable for the organization) is not a long
term project, but its effects are long term and must be planned carefully. First,
because the main driver of the change usually is cost savings (like other outsourcing
processes). Nevertheless, once the decision is made, the utility model will be there for
good. In other words: if the driver is cost savings, the opportunity window may be not
always the same. Furthermore, any outsourcing process implies a cultural change that
needs some time to mature (before and after the project kick off).
Roles: Though there are two clearly identified roles (service providers and clients),
the truth is that there are many others that participate and/or are affected by the
decision: Public Administrations, other companies and people. For instance, when a
European citizen entitles to a EU company to keep their personal data, but when the
company places its data in the cloud, the preservation of the rights is not clear.
Decision Variables: The first decision variable in this kind of decision making
problem is whether migrating the legacy is viable, either in its entirety or some
subsystems, or it's better to keep the system just as it is.
If the migration is proposed, another decision variable is which service and
deployment models we choose for implementing the decision. The service models
that we usually find in literature are: software as a service (SaaS), platform as a
service (PaaS) and infrastructure as as service (IaaS). Deployment models usually
proposed are: public cloud, community cloud, managed cloud, hybrid cloud and
private cloud. Each of the service models and deployment models has different
characteristics of technical implications, cost, scalability, security, propriety, location,
etc. and solutions proposed may be different for each subsystem.
Finally, a third decision variable is what is service level that the company wants to
agree with the cloud service provider for each of the clauses in the service level
agreement (SLA). It can be anticipated that the provider will charge differently for
each service level agreed in the SLA (availability, unattended management,
monitoring, etc.). This variable is, furthermore, quite complex because can imply
different combinations and/or technical developments.
95
Table 1. Deployment models.
Managed by Property of
infrastructure
Infrastructure
located in
Accessible and used
by
Public cloud
3
rd
Party provider 3
rd
Party provider Out of premises Non trusted
Community
cloud
3
rd
Party provider 3
rd
Party provider /
Community
In premises /
Out of premises
Trusted
Managed
cloud
3
rd
Party provider 3
rd
Party provider In premises Trusted
Private cloud
Organization / 3
rd
Party provider
Organization / 3
rd
Party provider
In premises Trusted
Hybrid cloud
Organization &
3
rd
Party provider
Organization &
3
rd
Party provider
In premises & out of
premises
Trusted & Non trusted
3 Multi-criteria
The first thing we can see when we look at the problem is the fact that we do not face
a situation in which a single goal can be optimized. We have several goals, some of
them conflicting each other, some of them difficult to quantify (therefore to optimize).
Some of the goals (i.e. reducing the maintenance costs and improving the system
availability) are easy to measure, but are at odds with each other. Others, like
improving intrinsic security are hard to measure.
Some works, like [19], have presented into quantifying some aspects of security.
But, when the main driver for the change is economical, the best path is to quantify
security in terms of its economic impact.
Formally, a MCDA problem can be expressed as:
()
(
)
(
){}
Aaag,,ag,agmax
k2
/
1
(3.1)
Where A is a finite set of possible alternatives
{
}
n2
a,,a,a
1
and
{
g
1
(
.
)
,g
2
(
.
)
,,g
k
(
.
)
}
a set of evaluation criteria.
Rarely an MCDA problem includes an alternative that optimizes all the criteria at
once. Furthermore, the solution is not only dependent on all input parameters in (3.1),
but on the beliefs of decision maker. The best solution is a compromise between the
given data and the preferences expressed by the decision makers.
In [12] Guitouni & Martel presented a methodology for a MCDA process
consisting in the non lineal recursive iteration with four steps: 1) structure the
problem, 2) model the preferences, 3) calculate the preference model, and 4) make
recommendations. The goal of the methodology was being able deal with all the
aspects of a decision problem in which a satisfactory solution is the new paradigm.
4 Legacy Systems Migration
The problem of low quality software from the perspective of maintainability and
adaptability of the systems is not new. Actually, this has been the core issue of the
whole practice of software engineering for many years. Whenever a new technology
appears, the products and services that were developed with the previous technologies
96
become legacy. For a given technology, the word ‘legacy’ is quite ambivalent: for the
owner of the system or for non IT executives ‘legacy systems’ are valuable assets,
whereas for the software practitioner of the IT executive they are a source of risks.
There are many definitions for ‘legacy system’, but probably the most extended in
the software engineering field are found in [22]: a system becomes legacy when
significantly resist modification and evolution regardless of the technology from
which it is built.” Some point out that legacy systems actually have a social part and
include, not only hardware and software, but processes and people. Another common
characteristic of this kind of systems is that they are classified as critical to business.
As for the methodology to evolve legacy systems we can find a framework in [3],
where Bisbal et al. classify the evolution of legacy from wrapping technologies to
overall redevelopment of the deprecated system, being common the mixed. Properly
speaking, migration is in between wrapping and redevelopment. In [22], Seacord et al.
also offers an approximation to modernizing driven by risk.
Nevertheless, most of the works we have found have one thing in common: they
are not specifically adapted to a given technology (that is, they have not been
tailored). There is little work on how to migrate to service oriented architectures
(SOA), work that should draw our attention since cloud computing actually is an
specific form of provision for this kind of architectures. In [13], Heckel et al., justify
the lack of work in the field of SOA since it’s an area for which concern for
reengineering is quite recent. Actually, if we set our focus into the more specific issue
of cloud computing, references are even less.
5 Security Criteria for Cloud Environments
It’s not surprising that there are plenty of research on security criteria for cloud
computing or SOA, either generic ones or quite specific.
Jansen & Grance’s report [15] is probably the most noticeable set of criteria and
specific issues related to cloud computing. In it, it is made a detailed description on
the criteria and issues related to privacy and security, identifying the pros and cons of
the model set up against its traditional counterpart, and the main points to cover.
They identify some positive aspects of the cloud model: specialization of the
technical people, greater uniformity leads to a greater security and eases automation,
resources availability, platform scalability, uniformity, robustness of the backup and
recovery policies. They also give some negative aspects (i.e. a greater system
complexity from the global perspective or the the loss of direct control either on the
logical or the physical items).
Since cloud computing steams from a set of different technologies like SOA, Web
2.0 and computing as an utility, many of the issues related to privacy and security
were already known. They are only viewed from a new perspective.
Jansen & Grance divide their security criteria in nine sections: security
governance, rules and standards compliance, trust building, security architecture,
identity and access management, software isolation, data protection, system’s
availability and incident response. For each of these sections, they make specific
recommendations.
Two other basic references for cloud security are [5] and [6] from the Cloud
97
Security Alliance (CSA). CSA presents its guide following three views and fifteen of
security: architecture, governance and operation. Each of the domains represent, to
our decision problem, a set of restrictions to our model that should be evaluated for
each of the legacy piece we have to deal with.
As we can see, none of this threats looks directly linked with the migration of
legacy. Nevertheless, the overall impression is that legacy migration does pose some
specific security questions.
6 Related works
There are plenty of works presenting decision models of some sort related to cloud
migration. One of them even presents a case. But most are not multi-criteria, and
nearly all focus solely on the aspect of expected economic savings.
Probably the most interesting one we have found is presented in [16, 17] which
studies specifically the use of decision support tools. Khajeh-Hosseini et al. present in
[17] two methods supported by tools to asses on the decision process on who is the
best provider and technology solution for cloud computing: though costs modeling,
and benefits and risks analysis. Costs modeling is presented as an extension to the
UML deployment diagram by means of allowing the modeler to incorporate items
like virtual machines or virtual storage, or an unspecific node, or remote applications
or data. In its risks/benefits analysis mode, [16] presents a more deep analysis
regarding the benefits and risks and taking into account diverse factors with an overall
decision outcome as a weighted sum of the risks and the benefits. Both, pros and
cons, are catalogued intro categories: technical, organizational, legal, security and
financial.
Though it’s not specific to MCDA, another interesting work is Schryen’s one in
[21]. In it, it’s presented a model based on fuzzy logic applied to the problem of
investing in security measures for distributed system models. Shryen proposes a
process that starting from a formal specification language for modeling security
conditions and its transformation into logical predicates, he defines some fuzzy sets
and a decision model based on those sets.
Another paper worth mentioning is Huang’s et al. [14]. In it, they use models like
DEMATEL, ANP and Grey Relational Analisys to address the analytical decision
process of baking services cloud migration.
There are other articles published on the issue of decision making for cloud
migration, some of them are closer to our problem (security), others less close:
Andrezejak, Kondo & Ji [1], Bibi, Katsaros & Bozanis [2], Chen & Sion [7],
Fedriksson & Agustsson [9], and Künsemöller & Kark [18]. We haven’t considered
them since, though they are related with MCDA techniques and cloud migration, they
do not deal with the migration concept in the sense that we talk about here.
7 Conclusions and Future Work
From the works we have seen up to now, we conclude that though there are quite a
98
few papers on the issue of MDCA applied to cloud computing decisions, there is
nothing specifically designed to address the problem of legacy system migration, and,
though some of them point out to security concerns none of them handles the security
issues as factors modeled in a quantitative manner integrated with the overall decision
model, but as constraints that must be ensured.
For that reason, we are working to develop an overall model to help the
practitioners to integrate security into de decision of migrating legacy systems to the
cloud with a MCDA perspective.
Given that any proposed model should take into account the actual beliefs and
preferences of the actual practitioners, we have developed a questionnaire to help us
to understand which are the true concerns of the IT decision makers who must face a
project to migrate part of their systems to the cloud, and how much security is a real
concern and how the value it against economic restrictions like costs.
The questionnaire will look into other issues like the anticipated and the actual
issues regarding the migration process, and the result of the projects (if they are
already finished).
We will feed this questionnaire to several IT executives and further refine our
model so that it can be used to future aid into the decision problem after modeling the
beliefs and preferences of the decision maker.
Acknowledgements
This research is part of the following projects: MEDUSAS (IDI-20090557) and
ORIGIN (IDI-2010043(1-5), financed by the Centre for Industrial Technological
Development (CDTI) and the FEDER, MAGO-PEGASO (TIN2009-13718-C02-01)
awarded by the Spanish Ministry for Science and Technology and SERENIDAD
(PEII11-0327-7035) and SISTEMAS (PII2I09-0150-3135) financed by the Council of
Education and Science of the Castilla-La Mancha Regional Government.
References
1. Andrzejak, A., D. Kondo, and S. Ji. Decision Model for Cloud Computing under SLA
Constraints. in IEEE International Symposium on Modeling, Analysis & Simulation of
Computer and Telecommunication Systems (MASCOTS), 2010. 2010.
2. Bibi, S., D. Katsaros, and P. Bozanis. Application Development. Fly to the Clouds or Stay
in-House. in 19th IEEE International Workshop on Enabling Technologies: Infrastructures
for Collaborative Enterprises (WETICE), 2010 2010.
3. Bisbal, J., D. Lawless, B. Wu, and J. Grimson, Legacy Information Systems: Issues and
Directions. IEEE Software, 1999. 16(5): p. 103-111.
4. Buyya, R., C. S. Yeo, S. Venugopal, J. Broberg, and I. Brandic, Cloud computing and
emerging IT platforms: Vision, hype, and reality for delivering computing as the 5th utility.
Future Generation Computer Systems, 2009. 25(6): p. 599-616.
5. CSA, Security Guidance for Critical Areas of Focus in Cloud Computing. 2009, Cloud
Security Alliance.
6. CSA, Top Threats to Cloud Computing. 2010, Cloud Security Alliance.
7. Chen, Y. and R. Sion. To Cloud Or Not To Cloud? Musings On Costs and Viability. in 2nd
99
ACM Symposium on Cloud Computing SOCC 2011. 2011.
8. Christiansen, C. A., C. J. Kolodgy, S. Hudson, and G. Pintal, Identity and Access
Management for Approaching Clouds, in IDC White Paper. 2010.
9. Fredriksson, J. and K. Augustsson, Cloud Service Analysis Choosing between an on-
premise resource and a cloud computing service. 2011, Chalmers University of
Technology, University of Gothenburg.
10. Gens, F. IT Cloud Services User Survey, pt.2: Top Benefits & Challenges. 2008; Available
from: http://blogs.idc.com/ie/?p=210.
11. Gomolski, B., Gartner Perspective on IT Spending 2010. 2010, The Gartner Group.
12. Guitouni, A. and J. M. Martel, Tentative guidelines to help choosing an appropriate MCDA
method. European Journal of Operational Research, 1998. 109(2): p. 501-521.
13. Heckel, R., R. Correia, C. M. P. Matos, M. El-Ramly, G. Koutsoukos, and L. F. Andrade,
Architectural Transformations: From Legacy to Three-Tier and Services. Software
Evolution, 2008: p. 139-170.
14. Huang, C.-Y., W.-C. Tzeng, G.-H. Tzeng, and M.-C. Yuan, Derivations of Information
Technology Strategies for Enabling the Cloud Based Banking Service by a Hybrid MADM
Framework. Smart Innovation, Systems and Technologies, 2011. 10: p. 123-134.
15. Jansen, W. and T. Grance, Guidelines on Security and Privacy in Cloud Computing. 2011,
NIST.
16. Khajeh-Hosseini, A., D. Greenwodd, and I. Sommerville. Cloud Migration: A Case Study
of Migrating an Enterprise IT System to IaaS. in IEEE 3rd Int. Conf. on Cloud Computing
(CLOUD 2010). 2010.
17. Khajeh-Hosseini, A., I. Sommerville, J. Bogaerts, and T. P. Decision Support Tools for
Cloud Migration in the Enterprise. in IEEE 4th Int. Conf. on Cloud Computing (CLOUD
2011). 2011.
18. Künsemöller, J. and H. Kark. A Game-Theoretical Approach to the Benefits of Cloud
Computing. in 8th Intl. Workshop on Economics of Grids, Clouds, Systems, and Services
(Gecon2011). 2011.
19. Madan, B. B., K. Goševa-Popstojanova, K. Vaidyanathan, and K. S. Trivedi, A method for
modeling and quantifying the security attributes of intrusion tolerant systems. Performance
Evaluation, 2004. 56 p. 167–186.
20. Mell, P. and T. Grance, The NIST Definition of Cloud Computing 2011, NIST.
21. Schryen, G. A Fuzzy Model for IT Security Investments. in Sicherheit 2010. 2011.
22. Seacord, R., D. Plakosh, and G. Lewis, Modernizing Legacy Systems: Software
Technologies, Engineering Processes, and Business Practices. 1st ed. Addison-Wesley
Professional. 2003.
23. The Open Group, The Open Group Cloud Computing Survey. 2011, The Open Group
24. Tobin, M. and B. Bass, Federal Application Modernization Road Trip: Express Lane or
Detour Ahead? 2011, Meritalk.
25. Winkler, V. J. R., Introduction to Cloud Computing and Security, in Securing the Cloud.
Cloud Computing Security. Techniques and Tactics., E. Syngress, Editor. 2011. p. 25.
100