2015 International Conference on Computational Intelligence & Networks

Cloud Computing Features, Issues and Challenges:

A Big Picture
Deepak Puthal∗ , B. P. S. Sahoo† , Sambit Mishra‡ , and Satyabrata Swain‡
∗ Dept.of Computer Science and Engineering, National Institute of Technology Silchar, India
† Dept. of Computer Science and Information Engineering, National Central University, Taiwan (R. O. C.)
‡ Dept. of Computer Science and Engineering, National Institute of Technology Rourkela, India
Email: {deepak.puthal, biswap.singh, skmishra.nitrkl, satya.swain10}@gmail.com

Abstract—Since the phenomenon of cloud computing was pro- in service demand [2]. Promoting and expanded expansion
posed, there is an unceasing interest for research across the globe. of cloud offers made a vast build-up around the cloud that
Cloud computing has been seen as unitary of the technology led to strong user expectation pressure, that partially couldn’t
that poses the next-generation computing revolution and rapidly
becomes the hottest topic in the field of IT. This fast move be reasonably be satisfied - this is by and large the case for
towards Cloud computing has fuelled concerns on a fundamental any promising technologies or concepts. Marketing tends to
point for the success of information systems, communication, guarantee attributes that are effortlessly confounded with qual-
virtualization, data availability and integrity, public auditing, ities with distinctive implications in different domains, pos-
scientific application, and information security. Therefore, cloud sibly prompting the confusion towards cloud. Consequently,
computing research has attracted tremendous interest in recent
years. In this paper, we aim to precise the current open challenges thorough characterization of cloud application features is an
and issues of Cloud computing. We have discussed the paper in essential for the further improvement of cloud framework.
three-fold: first we discuss the cloud computing architecture and Over the years, several technologies such as virtualization,
the numerous services it offered. Secondly we highlight several grid computing, and service-oriented architecture (SOA) have
security issues in cloud computing based on its service layer. Then matured and significantly contributed to making cloud comput-
we identify several open challenges from the Cloud computing
adoption perspective and its future implications. Finally, we ing viable [5]. On the other hand, cloud computing is still in in-
highlight the available platforms in the current era for cloud fancy and experiences absence of institutionalization in many
research and development. aspects. In current scenario, most new cloud providers propose
Keywords-Cloud computing; Cloud security; Virtualization; their own solutions and proprietary interfaces for access to
Workflow scheduling; Data integrity, Public auditing. resources and services which lead to the heterogeneity problem
and raises barriers to cloud realization. As users get more
I. I NTRODUCTION experienced in using cloud infrastructures, their capabilities,
In recent years, the popularity and rapid growth in process- strengths and deficiencies become more and more apparent.
ing and storage technologies and the success of the Internet, The cloud providers are thus working under growing pressure
computing resources have become cheaper, more powerful to fulfil the promises, and provide better services to their users.
and more ubiquitously available than ever before [1]. This As cloud infrastructure is being used throughout the globe,
technological trend is popularly known as cloud computing security is the major concern. This sharing of framework
and has led to an evolutive way to provide a better answer to together to the way that the customers to the cloud have needed
current and future information and communication technology control over the cloud foundation raises huge security worries.
(ICT) requirements. Cloud computing gives an adaptable on- The clouds have a different architecture based on the services
line environment which encourages the capability to handle an they provide. The information is stored on to a concentrated
expanded volume of work without affecting on the execution area called server farms having a huge size of information store
of the framework. With the advent of Cloud, the increasing and those data process in the server. So, the customers need
number of cloud providers and the variety of service offerings to trust the cloud resource provider on the accessibility and
have made it difficult for the researcher and pose numerous additionally information security. The service level agreements
challenges to cope with. Over the years, researchers are work- (SLA) is the only legal agreement between the service provider
ing around the world to enable this technology towards wide and client. The only means the supplier can addition trust of
business opportunity and in other areas of IT infrastructure, the client is through the SLA, so it must be institutionalize.
utilizing the cloud computing services and mechanism. Looking at the current trends and overgrowing interest for
Utilization of cloud services makes a developing relation- this subject, this paper explores the current patterns in the
ship among both public and private sector substances and the space of Cloud computing and presents researching space
people served by these elements. Cloud computing is attractive for future improvements of this technology. Key elements of
to business owners as it eliminates the requirement for users opportunity in cloud research are provided, and each one of
to plan ahead for provisioning, and allows enterprises to start them explained in detail.
from the small and increase resources only when there is a rise The rest of the paper is organized as follows. Section II

2375-5822/15 $31.00 © 2015 IEEE 114

116
DOI 10.1109/CINE.2015.31
presents a study on the cloud computing architecture and
highlights the recent available computing tools. In section III,
we list out several security issues in cloud computing. We
pinpoint the open challenges and discuss its future implication
in section IV. Finally in Section V, we conclude the paper.
II. C LOUD C OMPUTING A RCHITECTURE
A. The definitions
The term distributed computing appears to start from ma- Fig. 1. Cloud solutions based on the system’s deployment and service model.
chine system standard that speak to the web as a cloud. A
large portion of the real IT organizations and market research
firms such as IBM (2009), Sun Microsystems (2009), Gartner
by Plummer et al. [6], Forrester Research by Staten [7] and
Buyya et al. [9] have produced white-papers that effort to
define the meaning of cloud computing. These discussions
are basically reaching an end and a typical definition is
beginning to develop. The US National Institute of Standards
and Technology (NIST) has created a working definition that
blankets the generally concurred parts of cloud computing.
The NIST defines cloud computing as, a model for enabling
convenient, on-demand network access to a shared pool of con-
figurable computing resources (e.g., networks, servers, storage,
applications, and services) that can be rapidly provisioned
and released with minimal management effort or service
provider interaction [8]. This NIST definition describes cloud
computing as having five essential characteristics, three service
models, and four deployment models [2].
B. The essential characteristics of cloud computing [2] [8]
• On-demand self-service: Registering could resources be
gained and utilized whenever without the requirement
for human association with cloud administration sup-
pliers.Computing resources include processing power,
storage, virtual machines, etc.
• Broad network access: The beforehand said resources
could be gotten to over a system utilizing heterogeneous
gadgets, for example, laptops or mobiles telephones.
• Resource pooling: Cloud administration suppliers pool
their resources that are then imparted by numerous
clients. This is alluded to as multi-tenure where, for in-
stance, a physical server may have a few virtual machines
having a place with distinctive clients.
• Rapid elasticity: A client can rapidly gain more resources
from cloud by scaling out and can scale back in by dis- Fig. 2. Service model of Cloud: (a) Software as a service (SaaS), (b) Platform
charging those resources once they are no more needed. as a Service (PaaS), and (c) Infrastructure as a service (IaaS).
• Measured service: Resources utilization is measured by
monitoring storage usage, CPU hours, bandwidth usage,
etc. The said metrics are applied to all clouds, but each
cloud provides users with services at a different level of Software as a service (SaaS): Provides consumer the
abstraction, which is an alternate to an administration. capability to use applications running on a cloud infrastructure
C. The three most common service models of cloud computing mainly on the web browser to access software that offer as
a service over the web. The consumer do not have control
A cloud can collaborate with customer/client in a mixed bag
or figure out how to the underlying framework including
of courses, through capacities called services. Across the web,
system, servers, network, operating systems, storage, or even
three major types of models, of services have emerged, Fig. 2
individual application capacities, with the conceivable exemp-
shows the details of cloud computing service model.
tion of constrained client particular application setup settings.

117
115
GoogleDocs1 and Salesforces2 are prominent examples. specific requirements. This is typically for the shared concern
Platform as a Service (PaaS): Provides the capability to (e.g. such as schools within a university).
deploy onto the cloud infrastructure, consumer created appli- Hybrid cloud: A cloud that setup using a mixture of two or
cations, produced using set of programming languages and more private, public, or community cloud. In the hybrid cloud
tools that are supported by the Paas provider. The consumer could be freely overseen yet applications and information
does not oversee or control the underlying cloud framework would be permitted to move over the cloud.
including network, servers, operating systems, or storage,
yet has control over the sent applications and conceivably E. Features of the cloud computing
application facilitating environment arrangements. Much the The cloud is now hosting wide range of large scale and
same as the Saas model, clients do not have control or access small scale applications. Many organization or companies are
to the underlying base being utilized to have their applications now moving key applications from expensive internal data
at the Paas level.Examples of PaaS are Google App Engine3 centers to cost effective and resourceful cloud solutions.
and Microsoft Azure4 are prominent examples that use the Scalability: When a user lunch website scalability defines
PaaS model of cloud computing. a site or application’s skill to use traditional solutions on
Infrastructure as a service (IaaS): Provides the consumer demand. The site may scale up to available additional re-
with the capability to provision processing, storage, networks, sources the system is experiencing high user demand and later
and other fundamental computing resources from an IaaS may scale down recourse when the user demand turns down.
provider, and allow the consumer to deploy and run any Applications that run within the cloud are normally highly
software, which can include operating systems, services and scalable. An applicant can manually add or remove resources
applications. The client has control over operating systems, or application can be configured to scale automatically.
storage, deployed applications and perhaps constrained control Virtualizations: Virtualization is to use hardware or soft-
of select systems administration parts. Rather than the Paas ware to create the observation of something. Must server have
model, the Iaas model is a low level of reflection that permits their own CPU that is capable of running specific a specific
clients to the right of the entrance the underlying foundation operating system (OS), such as Windows, Linux, or Mac OS.
through the utilization of virtual machines. Iaas gives clients By using special software, server can be shown as it has
more adaptability than Paas as it permits the client to convey multiple CPUs and are running the same or different operating
any product stack on top of the operating system. Examples systems and the server CPU switches its processing power
of IaaS are Amazon Web Services EC2 and S35 are prominent frequently among the various operating systems.
examples that use the IaaS model of cloud computing. In the same way, desktop PCs typically run one operating
system. Again, by using special virtualization software, a desk-
D. The four deployment models of cloud computing
top PC/ laptop can be run simultaneously different operating
A cloud organization model indicates how resources inside systems. This provides an excellent platform for developer’s
the cloud and shared. Fig. 1, shows four different cloud application testers, and help desk support personal which
deployment models: private cloud, public cloud, community support multiple operation systems. without having multiple
cloud, and hybrid cloud. Each model impacts the comparing systems on the desk, the user can use multiple operation
scalability, reliability, security, and cost. systems in a single desktop PC.
Private cloud: A cloud that is used exclusively by one
organisation, company, or one of its customers. The cloud may F. Cloud Computing Simulators
be operated by himself or a third party, private cloud offers
During the study we compared various available cloud sim-
increased security at greater cost. The St. Andrews Cloud
ulators, their properties and unique features. The comparison
Computing Co-laboratory6 and Concur Technologies (Lemos,
study along with the research group working on these tools
2009) are illustration associations that have a private cloud.
are summarized in Table I.
Public cloud: A cloud that can be used by general public.
Due to its openness the cloud may be less secure. Public cloud III. S ECURITY I SSUES IN C LOUD
is the best option with less expensive. This can be a large or-
ganization and offer services. Public clouds require significant Here in this section we described several cloud computing
investment and are usually owned by large corporations such security issues based on different service layer. The Fig. 3
as Microsoft, Google or Amazon. shows the overlay architecture of security issues and trust
Community cloud: A cloud that is shared by two or more requirement in a top-down service model [23]. Trust basically
several organizations or company and is usually setup for their works in a top-down design, as every layer needs to trust
the layer instantly beneath it, and obliges a security ensure
1 http://docs.google.com at an operational, specialized, procedural and lawful level to
2 http://www.salesforce.com/uk/crm/products.jsp
empower secure correspondences. But the security is treated
3 http://code.google.com/appengine
4 http://www.microsoft.com/windowsazure/ as individually in each service layer. Trust could be seen as
5 http://aws.amazon.com/ a sequence from the end client to the application holder, who
6 http://www.cs.standrews.ac.uk/stacc thusly believes the provider.

118
116
 TABLE I
C OMPARISON OF CURRENTLY AVAILABLE CLOUD SIMULATORS

Simulator Base Platform Developer Available Language GUI Energy Model

CloudSim [30] SimJava University of Melbourne, Australia. Open Source Java No Yes
CloudAnalyst [31] CloudSim University of Melbourne, Australia. Open Source Java Yes Yes
iCanCloud [33] SIMCAN Universidad de Madrid, Spain. Open Source C++ Yes No
NetworkCloudSim [32] CloudSim University of Melbourne, Australia. Open Source Java No Yes
EMUSIM [34] CloudSim, AEF University of Melbourne, Australia. Open Source Java No Yes
GroudSim [35] - University of Innsbruck, Austria Open Source Java Limited No
MRCloudSim [36] CloudSim Seoul National University, South Korea Not available Java No Yes
DCSim [37] - University of Western Ontario, Canada. Open Source Java No No
SimIC [38] SimJava University of Derby, UK Not available Java No Rough
GreenCloud [39] NS2 University of Luxembourg, Luxembourg Open Source C++, otcel Limited Yes
MDCsim [40] CSIM Pennsylvania State University, USA Commercial Java, C++ No Rough
SPECI [41] SimKit University of Bristol, UK Open Source Java - Rough
MalStone [42] - University of Illinois, Chicago, USA Open Source Java, Python - Rough

A. Security issues in SaaS There are several highlights security issues in SaaS such
as data security, network security, data locality, data integrity,
In SaaS, the client needs to rely on upon the supplier for data segregation, data access, authentication and authorization.
fitting efforts to establish safety. The supplier must do the
B. Security issues in PaaS
work to keep numerous clients’ from seeing one another’s
information. So it gets to be hard to the client to guarantee that In PaaS, the administration supplier may give some control
right efforts to establish safety are set up furthermore hard to to the customer to manufacture applications on top of the
get confirmation that the application will be accessible when stage. However any securities beneath the application level,
required [24]. Based on SaaS, client can substitute net program for example, have and system interruption anticipation will
or software applications over old one. Hence, the center is not at present be in the extent of the supplier and the supplier
upon portability of uses, yet on safeguarding or upgrading brings to the table solid affirmations that the information stays
the security usefulness gave by the legacy application and distant between applications. Paas is proposed to empower
attaining effective information relocation [29]. designers to assemble their own particular applications on top
of the platform. As a result, it tends to be more extensible
The SaaS programming seller may have the application on
than SaaS, at the expense of customer-ready features. This
its own private server farm or convey it on a cloud computing
exchange off stretches out to security gimmicks and abilities,
framework administration gave by an outsider supplier (e.g.
where the implicit capacities are less finish, however there is
Amazon, Google, etc.). The utilization of cloud computing
more adaptability to layer on extra security [25].
coupled with the pay-as-you-go (develop) methodology helps
Applications sufficiently perplexing to influence an Enter-
the application administration supplier diminish the interest in
prise Service Bus(ESB) need to secure the ESB straightfor-
foundation benefits and empowers it to focus on giving better
wardly, leveraging a convention, for example, Web Service
administrations to clients. Over the past decade, computers
(WS) Security. The capability to portion ESBs is not accessible
have become widespread within enterprises while IT services
in PaaS situations. Measurements ought to be set up to survey
and computing has become a commodity. Enterprises today
the viability of the application security programs. Among
view data and business processes (transactions, records, pric-
the immediate application, security particular measurements
ing information, etc.) themselves as strategic and guard them
accessible are defencelessness scores and patch scope. These
with access control and compliance policies. However, in the
measurements can show the quality of application coding.
SaaS model, enterprise data is stored at the SaaS providers
Consideration ought to be paid to how malignant on-screen
data center, along with the data of other enterprises. Moreover,
characters respond to new cloud application architectures that
if the SaaS provider is leveraging a public cloud computing
the darkened application parts from their examination. Pro-
service, the enterprise data might be stored along with the
grammers are liable to the assault noticeable code, including
data of other unrelated SaaS applications. The cloud supplier
but not constrained to code running in the client connection.
may, also, imitate the information at numerous areas crosswise
They are prone to assault the foundation and perform extensive
over nations for the reasons of keeping up high accessibility.
black box testing. The vulnerabilities of cloud are connected
Most enterprises are acquainted with the conventional on-
with the web applications as well as vulnerabilities connected
reason model, where the information keeps on residing inside
with the machine-to-machine Service- Oriented Architecture
the endeavour limit, subject to their approaches. Therefore,
(SOA) applications, which are progressively being conveyed
there is a lot of inconvenience with the absence of control
in the cloud [25].
and information of how their information is put away and
secured in the SaaS model. There are strong concerns about C. Security issues in IaaS
data breaches, application vulnerabilities and availability that In IaaS, the developer has better control over the security the
can lead to financial and legal liabilities [25]. length of there should not any security gap in the virtualization

119
117
director. Likewise, however in principle virtual machines may IV. O PEN R ESEARCH C HALLENGES AND D IRECTIONS
have the capacity to address these issues yet in practice there A. Handling Uncertainties
are a lot of security issues [26]. The other element is the un-
wavering quality of the information that is put away inside the It has been demonstrated that one of the difficulties in
supplier’s equipment. Because of the developing virtualization provisioning cloud assets is uncertainty [10] [11]. Resource
of ”everything” in data society, holding a definitive control uncertainty emerges from a number of issues including client
over information to the holder of information paying little location, content type, noxious exercises, heterogeneity, and so
respect to its physical area will turn into a subject of most forth. At times, multimedia content delivery application may
extreme investment. To accomplish most extreme trust and confront with disappointment of resources or it may experi-
security on a cloud asset, a few procedures would need to be ence the ill effects of absence of sufficient resources. Consid-
connected [27]. The security obligations of both the supplier ering that bandwidth is one of the most important resources
and the client incredible contrast between cloud administration especially in multimedia applications, lack of bandwidth may
models. Amazons Elastic Compute Cloud (EC2) (Amazon, lead to huge degradation in QoS. In multimedia Application
2010) IaaS offering, as a case, incorporates merchant obli- the amount of required bandwidth is largely affected by three
gation regarding security up to the hypervisor, importance factors: 1) Media application bandwidth demand; 2) user’s
they can just address security controls, for example, physi- workload bandwidth demand; and 3) user’s location and device
cal security, natural security, and virtualization security. The type. These factors change rapidly and may result in bandwidth
client, thus, is in charge of the security controls that identify shortage at some points. For provisioning enough bandwidth
with the IT framework including the OS, applications and predicting these factors is critical:
information [29]. • Multimedia application network bandwidth demand: Pre-
dicting the size of approaching video frames relying upon
the transient history of the previously observed frames.
Based on the cloud deployment IaaS inclined to various • User’s workload bandwidth demand: Predicting user’s
security issues. Private cloud is more protected compared to workload (number of people requesting video) given the
a public cloud. The most important issue is to protect the history of previous requests.
physical infrastructure of data centers. It can be damage by • Multimedia application resource demand: Dynamically
any natural disaster or damage is acquired to the framework predicting and capturing the relationship between mul-
deliberately. Infrastructure doesnt mean the hardware where timedia application Qos targets, current hardware equip-
data is processed and stored, it also include the where it ment allotment and changes in client’s workload patterns.
is getting transmitted. In cloud environment data transmitted
By predicting above factors, the cloud resource provisioner
from the source to destination through large number of third
can allocate the appropriate amount of bandwidth. Some
party. So there is huge possibility that information could
methods [12] [13] [14] [15] for bandwidth and resource
be directed through an interloper’s foundation [25]. Despite
demand prediction are proposed in isolation, but these methods
the fact that cloud construction modeling is an extemporized
do not consider all of the above factors as part of single
engineering, the underlying advances continue as before. As
prediction process. This project will focus on both these
cloud services are available online, it builds over internet and
factors for an appropriate prediction of bandwidth. Moreover,
securities in web are postured by the cloud. It provides client
in media applications clouds providers have to deliver media
access resources over the internet whenever supplier dwells at
to users according to their SLA agreement with multimedia
distinctive area.
application providers. One of the important components of
SLA is that media frames must be provided in the user device
Regardless of the fact that gigantic measure of security is before specified deadlines to guarantee a constant display. For
placed set up in the cloud, still the information is transmitted addressing this issue, cloud providers must deal with link
through the ordinary underlying Internet. So threaten on the delays between clients and cloud servers. An important factor
Internet is leading to cloud threaten. But, in a cloud, the that directly affects the delay is the location of the media
dangers are devastatingly high. Cloud frameworks still use servers. According to the distribution of users appropriate
ordinary conventions and efforts to establish safety that are allocation of the servers, closer to the user, will result in lower
utilized within the Internet yet the prerequisites are at a higher delays. As the clouds provide possibility for allocating servers
degree. A dynamic set of arrangements and conventions are in different places, it is possible to address this issue by means
obliged to help secure transmission of information inside the of clouds more efficiently. This project will present methods
cloud. Encryption and secure conventions coddle the needs for proper allocation of clouds regarding the location of users.
to a certain degree yet they are not connection situated.
Concerns with respect to interruption of information by outer B. Handling dynamic variations in workload
nonclients of the cloud through the web ought to additionally An important benefit of hybrid architectures of cloud that
be considered. Measures should be set in place to make the has not been explored above, is their potential to handle
cloud environment secure, private and isolated on the Internet peaks in workload. In particular, the local data-center could be
to avoid cyber criminals attacking the cloud [25] [28]. provisioned with enough server ability to handle workloads,

120
118
 Fig. 3. Overlay architecture of security issues and trust requirement in a top-down service model.

while cloud assets could be invoked as required to manage D. Efficient of VMs Consolidation for Managing Heteroge-
peaks. The methodology could conceivably help in arranging neous Workloads
formats that can manage dynamic workload varieties. One
methodology is to utilize the model to focus the suitable Cloud infrastructure services provide users the capacity to
configurations for a mixed bag of evaluated workloads, and procurement virtual machines and assign any sort of uses
build the final configuration on the expected probabilities of on them. This prompts the way that distinctive sorts of uses
every workload. An alternate methodology is to utilize the (e.g., enterprise, scientific, and social network applications)
model occasionally as workloads change over time, to figure might be allotted on one physical machine node. On the other
out whether a change in placement is needed. More point by hand, it is not evident how these applications can impact one
point examination of these issues can be conceded and an open another, as they might be information, network or compute
opportunity to work on. intensive accordingly making variable or static load on the
resources. The issue is to figure out what sort of uses might
C. Optimisation of Virtual Network Topologies be designated to a single host that will give the most effective
In virtualised data centers VMs regularly communicate overall utilization of the resources. Current methodologies
between one another, making virtual system topologies. How- to energy proficient consolidation of VMs in data centers
ever, because of VM migrations or no streamlined alloca- do not investigate the issue of consolidating diverse sorts of
tion, the communicating VMs may wind up facilitated on workload. These methodologies typically concentrate on one
logically distant physical nodes giving expensive information specific workload type or do not consider various types of uses
exchange between one another. If the communicating VMs assuming uniform workload. In contrast to the previous work,
are dispensed to the hosts in diverse racks or enclosures, the an intelligent consolidation of VMs with different workload
network communication may include network switches that types can be proposed. A compute intensive (scientific) ap-
consume critical amount of power. To take out this infor- plication can be effectively combined with a web-application
mation exchange overhead and minimize power utilization, (file server) as the former mostly relies on CPU performance,
it is important to observe the communication between VMs whereas the latter utilizes disk storage and network bandwidth.
and spot them on the same or nearly placed nodes. To give In our opinion, which particular kind of applications can
compelling reallocations, the power utilization models can be be effectively combined and what parameters influence the
developed of the network devices and assessment the expense efficiency; and develop resource allocation algorithms for
of information exchange relying upon the traffic volume. As managing them can be investigated for a better approach.
relocations consume extra energy and they have a negative Moreover, this information might be applied to energy profi-
effect on the execution, before launching the relocation, the cient resource management strategies in data centers to attain
reallocation controller needs to guarantee that the expense of more ideal optimal allocation of resources and, accordingly,
migration does not surpass the profit. enhance usage of resources and decrease energy utilization.

121
119
For the resource suppliers, ideal distribution of VMs will A public verifier or a third-party auditor provide expert
bring about higher use of resources and, hence, diminished integrity checking services. During public auditing on cloud
operational expenses. End-users will profit from diminished information, the content of private information of an individual
costs for the resource utilization. client is not disclosed to any public verifiers. Hence, new
significant privacy issue, the leakage of identity privacy to
E. Scientific Workflow Scheduling public verifiers, is introduced. Failing to preserve identity
Cloud computing offers tremendous opportunities to solve privacy on shared data during public auditing will reveal
large-scale scientific problems in areas such as bioinformatics, significant confidential information to public verifiers. Devel-
astronomy, and physics. As cloud computing faced with var- oping techniques that can ensure the integrity of cloud storage,
ious challenges like performance variations and failures. The security and privacy remains a challenging research problem.
performance variations affect the overall execution time of the
G. Data Availability
workflow and failure affects the overall workflow execution
and increase the execution time. The workflow scheduling on Availability refers to the property of a framework being
distributed systems has been widely studied over the years and available and usable upon interest by an authorized entity.
is NP-hard by a reduction from the multiprocessor scheduling System availability refers to carry on operations even when
problem [17]. Current workflow scheduling on Clouds mostly authorities mischief. The system must have the capability to
focuses on homogeneous resources, and very fewer attempts carry on its operations even in the security violation. It also
have been made for the heterogeneous types of resources and refers data, software, and hardware available to authorized
one of the early attempts is made by Abrishami et al. [18]. The users based on the demand. The cloud likewise ensures that
workflow management systems should handle performance data and data handling is accessible to customers upon interest.
variations and failures while scheduling workflows. System availability incorporates a frameworks capacity to bear
There are two main stages when planning the execution of a on operations actually when some authorities mischief i.e. it
workflow in a cloud environment. In the resource provisioning must continue operations even during security break. Cloud
phase i.e. first stage, the computing resources are selected computing services show a substantial dependence on the
and provisioned. In the second stage, a schedule is generated resource frameworks and network accessibility at all times.
and each task is mapped onto the best-suited resource. The Comprehension and clearly documenting particular user
selection of the resources and mapping of the tasks is done requirements are imperative in planning a solution focusing at
so that different user defined quality of service (QoS) require- guaranteeing these requirements. Verifying identities many of
ments are met [19]. Previous works in this area, especially which impart basic crucial security necessities and deciding
those developed for Grids or Clusters, focused mostly on particular requirements for information protection and data
the scheduling phase. The reason behind this is that these security could be a standout amongst the most perplexing
environments provide a static pool of resources which are components of IS outline. This multiuser dispersed environ-
readily available to execute the tasks and whose configuration ment proposes exceptional security challenges, reliant on the
is known in advance. Since this is not the case in cloud envi- manager needs the level at which the user operates [23]. The
ronments, both problems need to be addressed and combined major security issues in the cloud system are as follows [4]:
in order to produce an efficient execution plan. Hence, the • Availability of information within participating systems;
future research efforts must take the dynamic provisioning • Maintain the integrity of information within the cloud,
and heterogeneity of unlimited computing resources in the i.e. preventing the loss or modification of information due
account to develop techniques and framework that cater the to unauthorized access or component failure. This is the
heterogeneous computing environment. major issue in public cloud;
• Provide control over access to services or their compo-
F. Public Auditing nents to ensure that all are authorized;
The popularity and rapid growth of cloud storage services • Clear separation of data and processes on the virtual level
to impart information to others has prompted an uncertainty in of the cloud, ensuring zero data leakage between different
the integrity of data in cloud storage, as data stored in the cloud applications.
can easily be lost or undermined because of the inescapable • To maintain the same level of security when adding or
hardware/software failures and human errors [20]. There nu- removing resources on the physical level.
merous traditional approach for checking data correctness. The Here we have line out seven different research directions in
conventional approaches are able to successfully check the the current era of cloud computing research. Based on the fea-
correctness of cloud data. However, the efficiency of using tures of cloud computing, we divided our research directions
this traditional approach on cloud data is in doubt, as it is into different aspects such as scalability, virtualization, data
required to retrieve the entire data from the cloud, and then management, cloud security and scientific applications. Those
verify data integrity by checking the correctness of signatures areas cover all the aspects of the cloud research and are helpful
of the entire data [21]. The mechanism that efficiently perform for a researcher to find out a unifying research direction. All
integrity checking without downloading the entire data from the research opportunities describe the concept/working model
the cloud is referred to as public auditing [22]. in-depth and show the direction of future research aspects.

122
120
 V. C ONCLUSION [19] Rodriguez, M.A; Buyya, R., ”Deadline Based Resource Provisioning
and Scheduling Algorithm for Scientific Workflows on Clouds,” Cloud
This paper discussed the emerging research issues that pur- Computing, IEEE T. on, 2(2), pp.222-235, 2014.
sued the advance scientific features of cloud computing with [20] K. Ren, C. Wang, and Q. Wang, Security Challenges for the Public
Cloud, IEEE Internet Computing, 16(1), pp. 69-73, 2012.
layer wise classification of the cloud services, and highlighted [21] Boyang Wang; Baochun Li; Hui Li, ”Oruta: privacy-preserving public
the subsequent guidelines of research facing the both indus- auditing for shared data in the cloud,” Cloud Computing, IEEE Transac-
try and academic community. This survey and future issues tions on , 2(1), pp.43-56, 2014.
[22] C. Wang, Q. Wang, K. Ren, and W. Lou, Privacy-Preserving Public
demonstrated that there are a few routes in which the cloud Auditing for Data Storage Security in Cloud Computing, Proc. IEEE
research group can gain from related groups. We have given an INFOCOM, pp. 525-533, 2010.
extensive outlook of current research issues cloud computing [23] Zissis, Dimitrios, and Dimitrios Lekkas. ”Addressing cloud computing
security issues.” Future Generation Computer Systems, 28(3), pp. 583-
and available platform to simulate the research idea. We have 592, 2012.
exhibited scientific classification of issues found here, and [24] Choudhary, V. Software as a service: Implications for investment in
the methodologies in which these issues have been handled, software development. 40th Annual Hawaii International Conference on
concentrating on an operational level, client level, service level System Sciences, IEEE, 209a-209a, 2007.
[25] Subashini, S. and V. Kavitha (2011). ”A survey on security issues in
and application level, security and context-awareness. service delivery models of cloud computing.” Journal of Network and
Computer Applications, 34(1), pp.1-11, 2011.
R EFERENCES [26] Gajek, S., et al. Breaking and fixing the inline approach. ACM workshop
on Secure web services, ACM, 2007.
[1] Armbrust, Michael, et al. ”A view of cloud computing.” Communications
[27] Descher, M., et al. Retaining data control to the client in infrastructure
of the ACM, 53(4), pp. 50-58, 2010.
clouds. International Conference on Availability, Reliability and Security,
[2] Sasikala, P. ”Research challenges and potential green technological appli-
IEEE, pp. 9-16, 2009.
cations in cloud computing.” International Journal of Cloud Computing,
[28] Staten, James, et al. ”Is cloud computing ready for the enterprise.”
2(1), pp. 1-19, 2013.
Forrester Research, 2008.
[3] Zissis, Dimitrios, Dimitrios Lekkas. ”Addressing cloud computing secu-
[29] Seccombe A, et al. Security guidance for critical areas of focus in cloud
rity issues.” Future Generation Computer Systems, 28(3), pp. 583-592,
computing. Cloud Security Alliance, 2(1), 2009.
2012.
[30] Calheiros, Rodrigo N., Rajiv Ranjan, Anton Beloglazov, Csar AF De
[4] R. Sherman, Distributed systems security, Computers & Security 11 (1),
Rose, and Rajkumar Buyya. ”CloudSim: a toolkit for modeling and
1992.
simulation of cloud computing environments and evaluation of resource
[5] Fernando, Niroshinie, Seng W. Loke, and Wenny Rahayu. ”Mobile cloud
provisioning algorithms.” Software: Practice and Experience, 41(1), pp.
computing: A survey.” Future Generation Computer Systems, 29(1), pp.
23-50, 2011.
84-106, 2013.
[31] B. Wickremasinghe, R. N. Calheiros, R. Buyya, CloudAnalyst: A
[6] Plummer, D.C., Bittman, T.J., Austin, T., Cearley, D.W. and Smith, D.M.
CloudSim-based Visual Modeller for analysing Cloud Computing En-
”Cloud Computing: Defining and Describing an Emerging Phenomenon.”
vironments and Applications, 24th IEEE International Conference on
Gartner, 2008.
Advanced Information Networking and Applications, 2010.
[7] Staten, J. ”Is Cloud Computing Ready for the Enterprise”, 2008.
[32] Garg, S. K., & Buyya, R. NetworkCloudSim: modelling parallel appli-
[8] Mell, P. and Grance, T. ”The NIST Definition of Cloud Computing.”
cations in cloud simulations. In Utility and Cloud Computing (UCC), 4th
2009.
IEEE International Conference on, pp. 105-113, 2011.
[9] Buyya, R., Yeo, C. and Venugopal, S. ”Market-oriented cloud computing:
[33] A. Nunez, et al. iCanCloud: A Flexible and Scalable Cloud Infrastructure
vision, hype, and reality for delivering IT services as computing utilities,
Simulator, Jr. of Grid Computing, 10(1), pp 185-209, 2012.
HPCC, 10th Proceedings IEEE, pp. 5-13, 2008.
[34] R. N. Calheiros, M .A. S. Netto, C. A. F. De Rose, and R. Buyya,
[10] Schad, Jrg, Jens Dittrich, and Jorge-Arnulfo Quian-Ruiz. ”Runtime
EMUSIM: an integrated emulation and simulation environment for mod-
measurements in the cloud: observing, analyzing, and reducing variance.”
eling, evaluation, and validation of performance of cloud computing
Proceedings of the VLDB Endowment, pp. 460-471, 2010.
applications, Software-Practice and Experience, 43(5), pp. 595-612, 2012.
[11] Iosup, Alexandru, Nezih Yigitbasi, and Dick Epema. ”On the per-
[35] S. Ostermann, K. Plankensteiner, R. Prodan, Th. Fahringer, GroudSim:
formance variability of production cloud services.” CCGrid, 2011 11th
An Event-Based Simulation Framework for Computational Grids and
IEEE/ACM International Symposium on, pp. 104-113. IEEE, 2011.
Clouds, Euro-Par 2010 Parallel Processing Workshops Lecture Notes in
[12] Di Niu; Zimu Liu; Baochun Li; Shuqiao Zhao; , ”Demand forecast and
Computer Science, pp. 305-313, 2011.
performance prediction in peer-assisted on-demand streaming systems,”
[36] J Jung,H Kim, ”MR-CloudSim: Designing and implementing MapRe-
INFOCOM, Proceedings IEEE , pp. 421-425, 2011.
duce computing model on CloudSim, International Conference on ICT
[13] Al-Tamimi, A.-K.; Jain, R.; So-In, C.;, ”Dynamic resource allocation
Convergence (ICTC), pp. 504-509, 2012.
based on online traffic prediction for video streams,” Internet Multimedia
[37] M. Tighe, G. Keller, M. Bauer, H .Lutfiyya, DCSim: A Data Centre Sim-
Services Architecture and Application(IMSAA), 2010 IEEE 4th Interna-
ulation Tool for Evaluating Dynamic Virtualized Resource Management,
tional Conference on, pp. 1-6, 2010.
8th international conference and 2012 workshop on systems virtualization
[14] Caron, E.; Desprez, F.; Muresan, A.; , ”Forecasting for Grid and Cloud
management (svm) Network and service management (cnsm), pp. 385-
Computing On-Demand Resources Based on Pattern Matching,” Cloud
392, 2012.
Computing Technology and Science (CloudCom), IEEE 2nd International
[38] S. Sotiriadis, N. Bessis, N. Antonopoulos, A. Anjum, SimIC: Designing
Conference on , pp. 456-463, 2010.
a new Inter-Cloud Simulation platform for integrating largescale resource
[15] Kalyvianaki, Evangelia, Themistoklis Charalambous, and Steven Hand.
management, IEEE 27th International Conference on Advanced Informa-
”Self-adaptive and self-configured CPU resource provisioning for virtual-
tion Networking and Applications, pp. 90-97, 2013.
ized servers using Kalman filters.” In Proceedings of the 6th international
[39] Kliazovich, Dzmitry, Pascal Bouvry, and Samee Ullah Khan. ”Green-
conference on Autonomic computing, ACM, pp. 117-126. 2009.
Cloud: a packet-level simulator of energy-aware cloud computing data
[16] Poola, Deepak, Saurabh Kumar Garg, Rajkumar Buyya, Yun Yang,
centers.” The Journal of Supercomputing, 62(3), pp. 1263-1283, 2012.
Kotagiri Ramamohanarao. ”Robust scheduling of scientific workflows
[40] Lim, Seung-Hwan et al. ”MDCSim: A multi-tier data center simulation,
with deadline and budget constraints in clouds.” 28th IEEE Int. Conf.
platform.” In Cluster Computing and Workshops, 2009. CLUSTER’09.
on Advanced Information Networking and Applications, pp. 1-8. 2014.
IEEE International Conference on, pp. 1-9. 2009.
[17] T. Sousa, A. Silva, and A. Neves, Particle swarm based data mining
[41] Sriram, Ilango. ”SPECI, a simulation tool exploring cloud-scale data
algorithms for classification tasks, Parallel Comput., 30(5), pp. 767783,
centres.” In Cloud Computing, pp. 381-392. Springer, 2009.
2004.
[42] Bennett, Collin, et al. ”Malstone: towards a benchmark for analytics on
[18] S. Abrishami, M. Naghibzadeh, and D.H.J. Epema. ”Deadline-
large data clouds.” In Proceedings of the 16th ACM SIGKDD Int. Conf.
constrained workflow scheduling algorithms for infrastructure as a service
on Knowledge discovery and data mining, pp. 145-152. 2010.
clouds.” Future Generation Computer Systems, 29(1), pp. 158-169, 2013.

123
121