![Disaster Recovery mode in the Cloud [16]](https://www.researchgate.net/profile/Hassen-Rebah/publication/318410948/figure/fig1/AS:969344448204802@1608121134273/Disaster-Recovery-mode-in-the-Cloud-16_Q320.jpg)
Content uploaded by Hassen Ben Rebah
Author content
All content in this area was uploaded by Hassen Ben Rebah on Dec 16, 2020
Content may be subject to copyright.
Hassen BEN REBAH
Laboratory SOIE
Higher Institute of Technological Studies
Mahdia, Tunisie
ben_rebah_h@yahoo.fr
Abstract— In a world of interdependent economies, the number
of threats imposed on the information system of companies and
organizations is steadily increasing. Companies need more than
ever to prepare for major crises and organize to be able to
ensure continuity of its critical IT activities through the
establishment of a Disaster Recovery Plan. For many
companies, the establishment of such plan was considered a
very difficult task to achieve. The development of cloud
computing offerings and their adoption within companies opens
up new practices. One of them is the rescue of the information
system in the cloud. In this paper, we focus on presenting this
new solution known as Disaster Recovery as a Service as well as
good practices in planning for a company to succeed such a
project. Our work was based on a comparative study between
traditional IT strategies and DRaaS solution and argued
through the enumeration of some research works as well as the
return to experience of some companies.
Keyswords — Disaster Recovery Plan, Traditional Disaster
Recovery Plan, Disaster Recovery as a Service, Cloud Computing.
I. I
NTRODUCTION
Today, all companies and organizations, whatever their
business activities, rely on computer tools and information
system to accelerate the process value chain. The conduct of
these processes is essential to the functioning of the
company [1]. In return , and despite its importance, the
computer tool within existing companies is still the weakest
link since the unavailability of an application or machine
room for a few hours may become a catastrophe that
generates harmful consequences as the big loss of revenue
or damage to brand image [2]. According to Philippe
Letreulle, technical director of Orbytes Engineering, US
studies have shown that, following the destruction of its IT
and telecoms means a company goes beyond a seventy-two
hours stoppage in 40% of cases1
1
. To anticipate such crises
companies must be prepared to ensure its business
continuity. Thus, the unavailability of a computerized
service is not a simple problem to solve. It is not enough to
save some electronic data carefully, plan some insurance
policies or replace the fire suppression system to believe that
the company is out of danger. It’s necessary to have a much
broader discussion and involve all company staff members
[3].
1
http://www.agence-c3m.com/communique/plan-de-reprise-dactivite-
informatique-orbytes-prend-le-relais-
Hatem BEN STA
Laboratory SOIE
Higher Institute of Management
Tunis, Tunisie
hatem.bensta@gmail.com
This holistic thinking implies that it is not just for the
company to take preventive measures to protect themselves
from threats affecting it but also to predict different
scenarios of crisis to maintain essential services and ensure
the recovery of its core activities through the establishment
of a Disaster Recovery Plan. However, the main obstacle to
the establishment of such a plan, in addition to its relative
logistical and technical complexity, is usually its cost [2][4].
The DRP is indeed an investment and a trickle charge in
significant operational condition, for use relating to
profitability, being potentially activated in case of crisis over
a disaster. The will to overcome these problems today leads
companies to turn to a new concept , cloud computing ,
characterized by a variety of IT service customized DRP
[5][6].
In this context, this work aimed to study the
shortcomings of the Traditional Disaster Recovery Plan
deployed previously in companies and the contribution of
the new disaster recovery strategy based on cloud
computing. We will start by exposing the Traditional
Disaster Recovery Plan as well as its shortcomings. Then,
we will introduce a new disaster recovery solution called
DRaaS based on the benefits of cloud computing while
emphasizing the contributions of this solution compared to
traditional recovery strategies. Then, we will focus on the
economic benefits of this disaster recovery model, the risks
that affected such a solution as well as the good practices
leading companies to successfully implement such a solution
within its premises. Finally, we will wrap up our paper with
a conclusion.
II. T
RADITIONAL
D
ISASTER
R
ECOVERY
P
LAN
(DRP)
A. What’s the DRP ?
According to CLUSIF 2004 “The DRP (Disaster
Recovery Plan) is a subcomponent of the BCP (Business
Continuity Plan) and it covers the IT resources. It guarantees
the revival of critical systems in a minimum set period. It
also guarantees the retrieval of data with a minimum of fixed
losses” [7].
On a disaster event, a company with a DRP must be able
to properly ensure the following activities
2
:
2
http://www.ssi-conseil.com/ssi-conseil2/www.ssi-
conseil.com/content/view/38/43/index.html
Disaster Recovery as a Service
A Disaster Recovery Plan in the Cloud for SMEs
• Turning into a Disaster Recovery Site.
• Switching the stricken site network links to the
Disaster Recovery site.
• Securing the recovery site network links.
• Restarting backup systems (servers).
• Restoring data ensuring their integrity.
• Checking the system availability and data across the
network.
• Restarting applications.
• Checking the functional availability of the
information system.
• Remaking the information system to the users.
These actions must be planned in accordance with two
time requirements defined in an interview conducted with
users of the information system:
• RTO (Recovery Time Objective): The targeted
duration of time in which business functions are
unavailable [8].
• RPO (Recovery Point Objective): The maximum
targeted period between two successive backups, and
thus the maximum amount of data that can be lost
due to a major incident when restoration is successful
[8].
B. Traditional Disaster Recovery strategies
In the past, to ensure their primary site, companies
generally use a second internal physical site with a second
computer room or external to a provider. This second site
can be classified according to emergency level:
• Hot site: it is a remote data center which is redundant
in accordance with the primary site.
• Warm site: is a remote computer room that is
partially equipped with computer equipments.
• Cold site: is a remote data center, available upon
activation of the DRP has no equipment installed in
advance. It is then necessary to supply new hardware
in case of disaster.
Once the type of disaster recovery site is set, it must be
supplied with the necessary data through various techniques:
• Asynchronous replication from one system to another
via an IP network.
• Synchronous replication of a disk array to another
array, via a storage area.
• The offset backup tapes to the disaster recovery site
through a means of transport.
C. Disabilities of traditional Disaster Recovery strategies
In 2012, Symantec "the world's leading in security
management solutions" has conducted a study of 200 French
SMEs on the planning and implementation of the DRP and
found that about 73 % of companies do not really have a
DRP
3
. This low rate of SMEs with a DRP can be explained
by several reasons; the most important one is related to
existing traditional disaster recovery strategies.
3
http://www.itespresso.fr/plan-reprise-activite-une-pme-sur-quatre-
elude-le-sujet-53781.html#tSUBQd4iBJZYG0CY.99
Generally, a company must plan its disaster recovery
strategy in accordance with RTO and RPO. These two time
requirements will place the company in either a dedicated or
shared disaster recovery model. In a dedicated model, the
infrastructure is dedicated to a single company. This type of
solution allows a shorter recovery time compared to other
conventional models; i.e. the IT infrastructure is mirrored on
the disaster recovery site and ready to be activated in case
of disaster. This process is dependent on the receipt of an
image updated data which involves carriage of physical tape
and a data restoration process. In addition, this strategy is
expensive because the equipment is not used in normal
times. Some companies use their backup infrastructure for
the purpose of testing and development to offset costs, but
this introduces an additional risk into the equation. Finally,
the process of restoring data induces an uncertainty in the
process. Data recovery can take up to 72 hours including
recovery time bands, transport and loading. In a shared
model, infrastructure is pooled among several companies. It
is supposed to be cheaper because the disaster recovery off-
site infrastructure is shared between multiple companies. In
case of incident, hardware, operating system and application
software on the disaster recovery site must be fully
configured to match the damaged site. This process can take
several hours or even days. In addition, it is necessary to
perform the data recovery process or an average of 48 to 72
hours before resuming [9].
Added to that, companies when developing and
implementing a DRP, face the following problems
4
:
• The difficulty to identify the perimeter of the DRP.
• The complexity of technical solutions to implement
whatever internally or externally.
• The mobilization of internal skills to operate the
DRP.
• The complexity and cost to maintain, over time, the
infrastructure of the two sites (primary and disaster
recovery site) at the same level.
• The difficulty in developing disaster recovery
solutions in terms of business requirements.
• The inability to organize regular testing of rescue.
These problems, mentioned above, make traditional
disaster recovery strategies often beyond the reach of SMEs,
mainly for financial reasons and complexity. When these
companies manage to overcome these obstacles, they are
struggling to adapt their DRP to the different changes in
their information system. This indicates a low confidence in
the chosen disaster recovery strategy.
III. D
ISASTER
R
ECOVERY AS A
S
ERVICE
:
DR
AA
S
A. Research work evaluating DRaaS performance
The disaster recovery via cloud infrastructure has been
addressed by several researchers who have proposed
models, approaches and techniques to enhance the
effectiveness of this universe in the area of company
4
http://www.scoop.it/t/le-monde-du-draas
business continuity. [10] gave a disaster recovery model for
web site application based on cloud. This model uses the
virtual machines in cloud computing to make data disaster
recovery service available. In this research, authors showed
that using cloud computing for data backup allows the
company to reduce the cost of data disaster recovery [11]
and to benefit from a very short recovery time in case of
disaster with an economic benefit of up to 85% compared to
an establishment of a traditional solution. [12] proposed a
comprehensive modeling and analysis of a Data Center with
Disaster Tolerance (DTDC). This work studied a typical HA
configuration of the DTDC consisting of two geographically
distributed data centers under the active/active pair. The
result of this study has shown that a data center with DT can
achieve a higher availability especially when located in an
area with less disastrous events. [13] provided an
architectural pattern describing the integration of high
availability and disaster recovery (HADR). This pattern for
database cluster replication combines simultaneously
synchronous and asynchronous replication concurrently for
high availability and disaster recovery purposes. In order to
evaluate the pattern and its simple implementation, we
simulated three DR scenarios for the commonly used DR
strategies hot standby, warm standby and cold standby, and
analyzed the availability, RPO and RTO, cost and response
time in terms of most SMEs concerns. The result of this
research showed that warm standby for DR is an ideal
strategy for most organizations in terms of the
comprehensive analysis of RPO, RTO and financial issues.
B. Why to adopt DRaaS solution?
As a hosted service in the cloud, DRaaS inherits the
following five characteristics of cloud computing model
5
:
• On-demand self service: the implementation of this
disaster recovery service will be fully automated.
• Broad network access: these datacenters are usually
connected directly to the Internet backbone to benefit
from excellent connectivity.
• Resource pooling: most of these centers have tens of
thousands of servers and storage means for fast
charge mounted.
• Rapid elasticity: the on-line new operational instance
of a server is achieved in a few minutes.
• Pay per use: the billing is calculated depending on
the duration and the amount of resources used.
C. Services that could be a recovery
Cloud computing offers three types of services that could
be helpful:
• IaaS (Infrastructure as a Service): is a model of cloud
computing that provides infrastructure components to
customers [14].
5
http://www.figer.com/Publications/nuage.htm#.VwqTqPnhDIV
• PaaS (Platform as a Service): is a model of cloud
computing that delivers application platform to the
costumer. [14].
• SaaS (Software as a Service): is a model of cloud
computing that provides on line software solution
[14].
D. Principle and operating mode of DRaaS
Recent studies by Gartner, Forrester Consulting and
Aberdeen Group, all show that CIOs have a strong passion
for these DRaaS technologies. According to Forrester
Consulting, more than half of US companies (59%) include
an implementation DRaaS project or are interested in this
topic
6
. Generally this solution is based on the use of the
virtual servers in the cloud which are started in case of
disaster. Therefore, users are working on cloud mode with
the saved data (i.e replicated and installed on these virtual
servers at the last automatic backup operation). When the
problem that caused the accident is technically solved, data
can be relocated to the company's servers and users can
connect as usual.
Figure 1. Disaster Recovery mode in the Cloud [16]
E. Different models of DRaaS
There are two types of DRaaS models
5
:
• The model "Do It Yourself" (DIY): it is the company
that chooses manages, builds and integrates the
various components needed to rescue its information
system.
• The service model: it is a third party (specialist
provider) that offers a complete solution with
integrated components in the cloud and offering a
guarantee of recovery.
6
http://ysosecure.com/PCA-PRA/draas-disaster-recovery-as-a-
service.html
F. Comparison between DRaaS and Tradionnal DRP
The analysis of the solutions (DRaaS and Traditional
DRP), previously effected, allows us to develop the
comparative table below:
TABLE I. COMPARISON
BETWEEN
DRAAS
AND
TRADITIONAL
DRP
Traditional DRP DRaaS
DR site
DR site is a secondary
physical site launched by the
company which is
responsible for its
preparation including
(installation of power,
cooling, backup servers,
network connectivity, etc)
A disaster recovery
service provider offers
company a virtual
machine snapshots of
their physical or virtual
servers lactated at the
primary site.
Manpower
The company must have the
necessary manpower
promoting the proper
complicated switching to the
DR site in case of disaster.
The DR administration
complexity is supported
by the provider.
Testing,
Maintenance
and Updating
Testing, maintenance and
Updating of DRP are
supported by the company
Testing, maintenance and
Updating of DRP are
supported by the provider
Access to the
DR site
Access to the DR site is
usually done via a cross-site
link already rented.
Access to the DR site is
usually done via Internet
from anywhere
Switchnig
time to the
DR site
The switching to the DR site
is usually done in a manual
way and it can vary from a
few hours to days
The switching to the DR
site is automatic and it is
of the order of a few
minutes
Cost Expensive Cheap
This comparison has shown that the DRaaS is much
more efficient and more flexible compared to a Traditional
DRP in terms of cost, rapidity and guarantee of data
recovery.
IV. E
XPERIENCE RETURN USING
DR
AAS
Following the study that was made in Section 3 and
found in the Table 1, the DRaaS solution has benefits far
greater than traditional DR. This will be implemented
further as follows:
A. Advantages of DRaaS economic model
The idea of DRaaS is to use the simplicity of the cloud
and its business model to provide DR solution. This
approach offers companies many advantages on the
technical, organizational or financial level. A study
conducted by Forrester Consulting in February 2012 with 75
leaders of DRP (US, UK, India) has shown many benefits
such as: [17]
• Pay per use: the DRaaS has a much better value
compared to Traditional DRP thanks to the pay per
use which is based on the consumed storage volume
instead of the inactive infrastructure. Servers, for
example, will cost the organization only in case of
rescue or test cases.
• Reduction of the DRP maintenance costs: DRaaS
solution reduces the administrative complexity of the
DRP for CIOs teams and thus reduces the costs
associated with the staff dedicated to maintenance
and updating of the DR solution.
• Facilitates access to tests: tests can be implemented
with limited cost regardless of the production. Thus,
the realization of the tests is encouraged by a lower
cost infrastructure, increasing the insurance of the
operational nature of the DR plan.
• Elasticity of the approach: the solutions of disaster
recovery in the cloud are flexible in terms of sizing
of necessary resources in the cloud. It is much easier
for CIOs teams to extend or reduce the scope of the
DRP on request from the provider without having to
deploy a new infrastructure or to modify the data
replication process.
• Best recovery time with a constant budget: DRaaS
type solutions easily ensure a resumption of
information system within hours in case of data
synchronization.
• Facilitates the automation of DRP: Under the
DRaaS, server activation and the switching process
can be automated upon the detection of a disaster to
allow the restoration of business services in minutes
without any intervention from CIO teams. This
instant automation generally benefits from a new
Network approach called, the New IP, which
combines new technologies such as the SDN and
network virtualization and is also based on a new and
a flexible Ethernet architecture "flat" called "Fabric
Ethernet
7
".
B. Risks related to the implementation of DRaaS
Risk analysis planned by LEXSI, the first service
company specialized in cyber security in France, has
identified eight major risks of implementing DRP in the
cloud. Some of these risks are shared with any project
related to the Cloud, such as: loss of data confidentiality;
others are specific to DRaaS issues, such as: the
impossibility of integration of specific architectures [16].
TABLE II. RISKS
RELATED
TO
DRAAS
Risk Scenario Consequences
R1 :
Inability to
integrate
specific
architectures
Poor estimation of project
difficulty can put it in
default.
- Project Failure
-Inability to implement
the DRASS for part of
the architecture.
R2 :
Loss of data
confidentiality
in the cloud
The causes are many: lack
of isolation environments,
usurpation, lack of data
encryption, etc.
- Image Degradation of
the company that may
have an economic
impact,
- Legal Proceedings,
- Confidential
information disclosure.
7
http://www.journaldunet.com/solutions/expert/49783/fabric--ethernet--
une-architecture-reseau-pour-les-datacenters-virtualises.shtml
R3 :
Breaking the
rules
Different rules may apply
to data hosted in the
DRaaS provider. In case
of breaking these rules,
legal and financial
penalties may be imposed.
- Legal Sanctions,
- Image degradation
- Loss of contract.
R4 :
provider's
inability to
ensure the
service
The unavailability of the
service provider can be
from a bankruptcy, unable
to rescue several
companies simultaneously
by lack of experience or
shared resources, an
inability to return due to
loss of service in case of
large-scale disaster , etc.
- Impossible rescue
- Potential Bankruptcy
of the company
R5 :
Legal disputes
related to non-
conformities
Legal disputes may be due
to:
- Data hosted in Cloud
within the countries in
which protection and data
processing is not
appropriate
- A loss of traceability of
data in the cloud.
-Legal Proceedings,
causing financial loss
and image degradation.
R6 :
Loss of data or
integrity of
replicated data
in the cloud
Operational problems of
the cloud, errors in the
data backup mechanism,
malicious people with
higher privileges in terms
of administration etc.: All
this can cause data loss.
- Serious
disorganization is due to
the loss of integrated
detected belatedly,
- Inability to rescue all
or part of information
system,
- Errors in the data /
Rescued systems.
R7 :
Inability to
recover data
This may be related to the
risk R3: If the provider is
unavailable, it is
impossible to access to the
data. This threat can also
be achieved in the context
of a conflict with the
service provider.
Data Loss
R8 :
Loss of
governance
There is frequently an
uncontrolled stack of
subcontractors which
causes a loss of control for
the customer
- Loss of control in case
of the lack of visibility
on data management
- Loss of governance
involving loss of the
recovery control.
C. Feed-Back: How to make DRaas project successful?
To succeed DRaaS project, the companies have to apply
an approach based on 21 points grouped in 7 topics [16]:
1) Topic 1: integration
a) Audit of existing: some architectures and some
applications cannot be considered in a DRaaS strategy.
Above all, start by detailing your architecture to the service
provider to assess whether its DRaaS solution is usable on
your information system or not. Beware of a service
provider who would ensure that the DRaaS is adaptable to
everything without seeking to know your particular
architecture.
b) Integration testing to supply and associate pricing:
tests need to light the standby servers in the cloud, make
sure that the pricing of tests is acceptable and no false
charge is concealed.
c) Adaptation of existing infrastructure: some providers
require some level of virtualization of the primary site from
their customers. Orientate these requirements because
providers work with different technologies. It is usually
possible to find providers adapting to any standard
conventional technologies.
d) Compatibility of the network throughput of the
different sites: the volume of data exchanged with the
DRaaS induces the need to increase the reliability and the
width of the bandwidth.
e) Software integration problem: the use of software
involves the participation of more staff members in the
implementation of an offer to DRaaS: the software provider,
DRaaS provider and the customer.
2) Topic 2: compliance
f) Compliance to customer requirements: you must
ensure compliance of the data and the effective respect of
national rules of compliance by the provider.
g) Compliance of service provider: you can ask your
provider or require compliance with certain standards (ISO,
PCI-DSS, SAS-70 Type II, etc.).
3) Topic 3: confidentiality
h) Hosting Zone: ask about accommodations and the
laws governing them to verify the absence of any
contradiction and contractually ensure that your data will
remain well in a particular country or geographic area.
i) Cloud model: the different cloud models must be
carefully analyzed (private, public and hybrid) depending on
the criticality of the resource, cost and regulatory demands
on the data hosted in the cloud.
4) Topic 4: availability
j) Continuity of service: ask specifically about the
status of the provider finances, experiences or references as
well as the level of training and experience of its staff.
k) Service Level Agreement: penalties of the provider in
case of non-compliance must be carefully defined.
l) Management of rescue simultaneous customers: it is
the customer's right to ask to know the distribution of
customers on data center of the provider to ensure recovery
in case of simultaneous recovery needs.
m) Reversibility plan: set a reversibility plan for the
guarantee to recover all data in the cloud.
n) Distance from hosting zone: the hosting zone of the
cloud should be chosen far enough away from the primary
site to ensure its availability in the extended major disaster
event.
5) Topic 5: configuration
o) Possibility of using DRaaS: you have to finish with
the provider all the possibilities of using DRaaS in order to
appropriately configure the Cloud.
6) Topic 6: security
p) Default Security Level: never accept offers black
box! Also, you should never consider that security is
included by default. The more you are informed, the more
you will master the security level.
q) Logical and physical security: ask the provider to
perform regular security audits to test both the physical and
logical security infrastructure.
r) Securing access to rescued Information System in the
cloud: access to information system rescued is by internet.
Check the type of authentication, encryption and traceability
to prevent any intrusion problematic during rescued mode.
s) Security of data transferred and exchanged flows:
security must not be neglected for the transfer of data from
the company's information system to the cloud. Make sure to
minimum encryption of data transferred.
7) Topic 7: strategy
t) Transparency of the provider in its internal strategy:
It is legitimate to know the strategy and service management
mode. Management based on ITIL v3 or ISO 20000
certification may be evidence of mastery of the service.
Nevertheless, it is always good to check that the provider
integrates perfectly security basics (RSSI, Charter Security
Policy, etc.).
u) Transparency of provider on security incidents: you
can impose transparency on other customers who would
share the same infrastructure like you and the incident
management strategy and privacy.
V. C
ONCLUSION
In this paper, we treated, on the one hand, the emergency
issue of information system disaster for SME as well as the
shortcomings of traditional solutions. On the other hand, we
have also proposed a new solution called "DRaaS" which
benefits from the main advantages of Cloud and offers SME
more beneficial solutions through improved quality recovery
service and a significant decrease of costs due to low prices
of data storage and pay per use. The effectiveness of this
solution was also justified in this work by listing several
researches offering DR models in the cloud and the feedback
from several companies.
Cloud computing is an economic and technological
revolution that is being increasingly imposed on SMEs
especially in terms of data storage and disaster rocovery. But
there's a question: Is really using a third party to save my
data an effective solution to protect myself?
R
EFERENCES
[1] L. Pejsachowicz, N.Thoré, S.Ruelle, T.Autret, F. Tête, V. Minassian
and J.M.GREMY. Plan de continuité d’activité plan de reprise
d’activité, Synthèse de la conférence thématique du CLUSIF, 2014.
[2] O.H.Alhazmi and K.M.Yashwant "Assessing Disaster Recovery
Alternatives: On-site, Colocation or Cloud," Software Reliability
Engineering Workshops (ISSREW), IEEE 23rd International
Symposium on. 2012.
[3] P.Boulet. "Plan de continuité d’activité, secours du système
d’information", Cachan : Hermes Science, France, 2008.
[4] S.M. Hawkins, D.C. Ye and D.C. Chou, "Disaster recovery planning:
a strategy for data security, Information Management & Computer
Security", Vol. 8, 2000, N°5, pp.222 – 230.
[5] M.A.Khoshkholghi, A.A.Rohaya Latip, S.Subramaniam and
M.Othman, "Disaster Recovery in Cloud Computing: A survey",
Computer and Information Science, Vol.7, 2014, N°4.
[6] M. Wiboonratr and K.i Kosavisutte, "Optimal strategic decision for
disaster recovery," Int. Journal of Management Science and
Engineering Management, Vol. 4 ,2009, No. 4, pp. 260-269.
[7] Lexique structuré de la continuité d’activité, Business continuity
structured glossar, Livre blanc n°1 du club de continuité d’activité
CCA, version 3.0, 2012.
[8] E.Besluau, "Management de la continuité d’activité, Assurer la
pérennité de l’entreprise : Planification, choix techniques et mise en
œuvre", Paris : Eyrolles, 2008.
[9] Cloud Computing : Reprise après incident en mode virtualisé, Pour
une transition rapide de vos applications vers un Cloud résilient, Livre
blanc pour l’innovation, IBM Global Technology Services, 2012.
[10] T. Wood, E. Cecchet, K. K. Ramakrishnan, P. Shenoy, J. Van der
Merwe, and A. Venkataramani, "Disaster recovery as a cloud service:
Economic benefits & deployment challenges", in 2nd USENIX
Workshop on Hot Topics in Cloud Computing, Boston, USA, 2010.
[11] Y.Gu, D. Wang, and C. Liu, "DR-Cloud: Multi-cloud based disaster
recovery service", Tsinghua Science and Technology, vol. 19, 2014,
No 1, pp. 13-23.
[12] T. A.Nguyen, D. S. Kim and J.S.Park," Availability modeling and
analysis of a data center for disaster tolerance", Future Generation
Computer Systems, vol.56,2016, pp.27-50
[13] H.Xiong, F.Fowley, C.Pahl, An Architecture Pattern for Multi-Cloud
High Availability and Disaster Recovery, Workshop on Federated
Cloud Networking FedCloudNet'2015, 2015.
[14] J. Chaudhary, V.Maheshwari, Proposed Architecture for Virtual Lab
Using Amazon Services, Int. Journal of Advanced Research in
Computer Science and Software Engineering, 5 (5),2015,pp.1637-
1640.
[15] Tout ce que vous devez savoir sur l’informatique dans le nuage, Livre
blanc du Cloud Computing, Syntec Informatique, 2010.
[16] Le secours du SI dans le Cloud, Faut-il faire le grand saut du DRaaS ?,
Livre blanc, Lexsi, 2014.
[17] PRA dans le Cloud pour les PME, DRaaS : Disaster Recovery as a
Service, Livre blanc, Nuabee, 2015.
