Innovation
We presented attractive approaches from various aspects to model and analyse complex data. We have
developed new frameworks, methodologies, and techniques for modelling and analysing complex data in
tobacco control. This methodology includes information technology, mathematical analysis, data mining, and
optimisation tools
We used our research outcomes to model and analyse complex data provided by the VicHealth
Centre of Tobacco Control, The Cancer Council Victoria. The data mainly come from the International
Tobacco Control Policy Evaluation Survey (ITCPES). In particular, we connectour research
171
outcomes for evaluation of the effects of existing instruments on smoking forbetter understanding
their roles and limitations, so that we can better identify where new actions might be required.
Despite of challenges and limitations, the research outcomes providenew insights into how to
maximize the effectiveness of Public Health policies. The research outcome fixed the gap between existing
techniques limitations and expected outcomes by providing a good and solid template for complex data in
tobacco control systems, as well as in other similar complex social data domain.
3. Learning environment design, and brain data sets. Brain Complex Networks
Discussions on “Brain Data sets” I would like to connect to the problem on:
“HOW BRAIN BEST LEARNS”
The process of learning has strong correlation with the brain functioning that is a very special area of
research in Neuroscience.
Brain networks as information processing systems
Background: The era of discovery science for human brain function was inaugurated by the
collaborative launch of the 1000 Functional Connectomes Project (FCP) on December 11, 2009. FCP
entailed the aggregation and public release (via www.nitrc.org) of over 1200 resting state fMRI (R-fMRI)
datasets collected from 33 sites around the world. In just over 6 months, the release generated over 9000
downloads and ~32,000 page-views from 1,223 cities in 78 countries.
1,000 Functional Connectomes Project is a collection of fMRI data sets donated by researchers from
35 centres around the world. This freely available resource includes data from more than 1,400 healthy
subjects who underwent fMRI scans that assessed their brain activity when their minds were at rest (Proc.
Natl. Acad. Sci. USA 107, 4734–4739, 2010). The study showed that resting-state fMRI data— long thought
of as nothing more than random, background noise—can be reliably pooled across scanners to unveil a
universal architecture of activity connections within the brain.
Unlike task-based fMRI, which can be highly specific to the study site, the 1,000 Functional
Connectomes resource allows for systematic explorations of healthy and diseased brains to discover hitherto
unknown underlyingdifferences. “We’re moving in the direction of being able to have objective measures of
neurological and psychiatric illness,” Milham says. “It’s all stepping in the direction of being a clinical tool.”
The effort takes its name from the Human Connectome Project, a $30 million initiative launched by
the US National Institutes of Health last year to map the entire physical circuitry of the healthy adult human
brain. But functional connectivity and structural connectivity is not the same thing. Functional connections,
for example, can span more than one synapse and can be modulated by emotion or sleep, whereas anatomical
circuits are more or less fixed over the short term.“Having this much data in one place is a real treasure trove
that is free to anybody who wants to play with it,” says Marcus Raichle, a pioneer of resting-state fMRI at
Washington University in St. Louis, Missouri who was not involved with the study.“The connectomes
project has the power to ask more questions,” adds Craig Bennett, a cognitive neuroscientist at the University
of California–Santa Barbara who published a review this month questioning the reliability and repeatability
of fMRI scans in most typical neuroimaging studies (Ann. N.Y. Acad. Sci. 1191,133–155, 2010). “You’re not
just looking across one study, you’re drawing from such a large body of research that you really say things
with authority.”
Since having been postedonline last December, the data set has been downloaded more than 4,500
times from researchers across 54 countries, according to Milham. One person who has explored the resource
is Nora Volkow, director of the US National Institute on Drug Abuse in Bethesda, Maryland. Volkow is now
developing quantitative methods to measure functional connectivity in her lab to follow up on preliminary
observations of systemic differences between males and females.
172
REFERENCES
1. Z. Dzalilov, A. Bagirov and M. Mammadov. Application of Optimization Based Data Mining Techniques to
Medical Data Sets: A Comparative Analysis, IMMM 2012, Proceedings of The Second International Conference in
Information Mining and Management, October, Venice, Italy, P: 41 to 46: ISBN: 978-1-61208-227-1;
2. Z. Dzalilov and A. Bagirov (2010). Cluster Analysis of a Tobacco Control Data Set. International Journal of
Lean Thinking.1(2): 40-5.
3. Z. Dzalilov, J. Zhang, A. Bagirov and M. Mammadov (2010). Application of optimisation–based data
mining technique to tobacco control dataset. International Journal of Lean Thinking.1(1):27-41.
4. G. Hafen, C. Hurst, J. Yearwood, M. Mammadov, J. Smith, Z. Dzalilov, P. Robinson. A new clinical
scoring system in Cystic Fibrosis: Statistical tools for database analysis-a preliminary report. BMC Medical Informatics
and Decision Making, 8: 44.
5. M.A. Mammadov, Rubinov A.M. and Yearwood, J. (2007), The study of drug-reaction relationships using
global optimization techniques. Optimization Methods and Software, Volume 22, No: 1, 99-126.
6. M. Zarei and Z. Dzalilov (2009). Optimization of back-propagation neural networks architecture and
parameters with a hybrid PSO/SA approach. Proceedings of fifth International Conference on Soft Computing,
Computing with Words and Perceptions in System Analysis, Decision and Control (ICSCCW 2009). Famagusta, North
Cyprus.
7. Z. Dzalilov, A. Bagirov and M. Mammadov. Application of Optimization Based Data Mining Techniques to
Medical Data Sets: A Comparative Analysis, IMMM 2012, Proceedings of The Second International Conference in
Information Mining and Management, October, Venice, Italy, P: 41 to 46: ISBN: 978-1-61208-227-1;
UDCI 004.056
Imanbayeva A. K.
1
, BissarinovB.Zh.
2
, Bissarinova A.T.
3
1
Kazakhstan, Kazakh National University named by Al-Farabi,
2
Kazakh National Technical University named after K. Satpayev
Almaty, Kazakhstan
bbaituma@gmail.com,
SECURITY IN THE CLOUD – VULNERABILITIES, THREATS, AND RESPONSES
Abstract. Cloud technology is becoming one of the fastest growing sectors of the IT industry due to the reduction
of costs on computation processes, along with benefits such as flexibility and scalability. Cloud computing is used
widely among a lot of organizations. However, this new technology opens new prospects for threats against security of
data. Mostly, threats in the cloud are similar to the regular attacks such as spyware, malware for data stealing, Trojans,
viruses, worms, bots and so on. Besides regular type of attacks, there are other issues associated with cloud due to the
infrastructure of the technology. This paper will discuss problems regarding reliability and security in the field of
virtualization and cloud computing, it will also propose available solutions to those problems. Real examples of cyber
attacks in cloud computing environment will be presented.
Key words: cloud computing, virtualization, security, cyber attacks.
Introduction
An environment with network infrastructure that is used for sharing data and computations is called
cloud computing. Clouds work based on the Internet and their purpose is to hide the complexity from users.
The notion of cloud computing includes both the equipment and software in data centers for provision of
services, and those services in the form of applications. Virtualization technologies are used for computations
on the cloud.
Currently Public, Private, and Hybrid cloud environments exist. In public cloud model resources can be
accesses by the public. Services provided by public cloud might or might not be charged. Private cloud
model services are internal to companies and are not available for the public use. If a part of resources are
managed by company internally and the rest is available for ordinary people then such kind of environment
is called hybrid cloud. The private part of a hybrid cloud is defended by firewalls and only authorized staff
has a permission to access it.
The services provided in the cloud can be divided into three major categories: SaaS (Software-as-a-
Service), PaaS (Platform-as-a-Service), and laaS (Infrastructure-as-a-Service) [2].
According to the results of the survey conducted on the global scale by Japanese security software-
company called Trend Micro, over a half of organizations who took part in the survey showed that they
173
utilize cloud technology. 45 percent of surveyed companies indicated that they utilize private cloud, whereas
46 percent seem to use private cloud (see Table 1) [1].
Table 1 – Implementation of virtualization and cloud computing
% that have currently deployed or are
piloting
T
ot
al
U
S
Ja
pa
n
Indi
a
G
erm
any
U
K
Ca
na
da
Server Virtualization
59
70
58
51
61
68
47
VDI
52
62
42
48
55
63
45
Public Cloud
45
54
37
38
48
52
42
Private Cloud
46
56
34
42
54
51
38
To use cloud infrastructure a lot of enterprises in a rush deploy simply physical server security on
virtual machines, but new security threats specific to cloud computing and virtualization are not considered
by typical physical server security. Furthermore, such kind of security might have negative influence on
platform performance.
Virtualization Security Threats
In this section threats and issues specific to virtualization infrastructure will be discussed.
Communication Blind Spots
Connections between virtual machines on the same host are not seen by conventional network security
tools. If outside the host machine all communications are connected to that security tool, then the
connections are visible. However, this security technique leads to time delays. Placing a special security
virtual machine on a host that can accord communication between other virtual machines can help to get rid
of invisibility and to decrease the amount of delay [3].
In a virtual environment this can be counted as a good solution. For cloud environment a special
security virtual machine is not ideal though, because such virtual machine should use hypervisor and in some
cloud environments hypervisor is not accessible. The virtual machines in the cloud are self-defending, thus
outside communication is not necessary.
Inter-VM attacks and hypervisor compromises
Operating systems and applications used by virtualized and physical servers are the same. Therefore,
attackers might use vulnerabilities of those applications and systems, and thus become a threat for virtual
environment. If attacker is able to compromise any one part of the virtualized environment, the other parts
are under threat as well, unless virtualization-aware security is provided [1].
Figure 1 – Inter-VM Attack
174
One scenario suggests that after compromise of one guest VM by an attacker, the compromised VM can
distribute the infection to other guest VMs on the same host. Close allocation of several VMs lifts the
chances of further compromise distribution. In this case, malware should be discovered by intrusion
detection and prevention along with firewall systems, without regard to the placement of the VM inside the
virtual environment.
Attackers also include hypervisor in their attack plans. Hypervisor is a program with a help of which
several VMs are able to run on a one computer. So, on the one hand hypervisors are a great help and on the
other hand it might lead to computing risks. That is why to have a secure hypervisor is very important task.
“Hyperjacking” is a type of attack when malicious software that entered one VM is able to attack the
hypervisor. Guest VM attacks a hypervisor, other VMs on that host are attacked by compromised hypervisor
[4].
To make a requests to the hypervisor VMs use different kinds of techniques, those methods usually tend
to have some API (application programming interface. The primary goal for the creation of API is to be able
to control VMs remotely from the host [5]. So, APIs are often attacked by malware. Therefore, APIs must be
secure and VMs should make only authorized requests.
Mixed Trust Level VMs
When the same host is occupied by mission-critical data VMs and VMs with less critical information
mixed trust level VMs are formed. Some companies may try to separate this secure data of mixed levels on
different host machines. However, this may result in thwarting of the aim of virtualized environment - to use
resources more effectively. For companies it is crucial that while the advantages of virtualization are being
realized, mission-critical data is safe. VMs can be protected with the help of self-defending VM security
even in environments of different trust levels. The protection tools include “detection and prevention, a
firewall, integrity monitoring, log inspection and antivirus capabilities”.
Instant-on-gaps
Even though virtualized environments are innately safer than their physical analogues, in practice
virtualization may pose a threat of having vulnerabilities, unless administrators know about them and take
certain actions to remove them. One of the possible vulnerabilities that may occur are instant-on gaps.
Companies use VMs in their needs to consolidate servers, decommission, migrate and clone VMs for testing
environments and VMs' dynamic nature is especially advantageous there [6]. Therefore, activating and
deactivating VMs, updating and securing them may be difficult.
After some time inactive VMs may diverge from the minimum security state that far that even
activating them may cause an occurrence of serious vulnerabilities in security. For example, some inactive
VMs may still be accessed by attackers even if they are inactive. Furthermore, security out of date might
facilitate cloning process of new VMs from templates.
Outdated security of VMs may enable attackers to maximize the benefit of using VMs for a longer time.
In general, when antivirus is being used or updated but guest VM is not online, the VM will become inactive
and unprotected. However, when a guest VM becomes online, it will become immediately vulnerable. A
solution for this problem could be a special security VM for every host which will update VMs automatically
when it is powered on or cloned. This gives companies an opportunity to realize advantages of virtualization
[1].
Cloud Computing Control And Security
Cloud computing is a result of addition of automation and virtualization. By the use of virtual
environments the capacity of physical servers is used to the full extent and thus contributes to the acquisition
of more computing power. It was discovered by service providers that by using virtualization it became
possible to enable multi-tenant usage of physical servers instead of single-tenant. Private clouds built on the
virtual infrastructure also seem to have improvements in utilization of resources and facilitation of service
supply. Different cloud models mentioned above (private cloud, public cloud and hybrid cloud) enable
distinct control levels and they differently affect security[1].
Cloud Computing Threats
Since cloud computing works based on virtualization threats discussed for virtual environments are also
dangerous for cloud computing. The boundaries of cloud computing covers a lot: information on public
clouds, private clouds and mobile devices. This opens new prospects for threats, and therefore accordingly
new security measures should be taken.
Security threats in cloud are cloning and rapid resource pooling, motility of data and data remnants,
175
elastic perimeter, unencrypted data, shared multi-tenant environments of the public cloud, control and
availability.
Cloning and rapid resource pooling
Regardless of the model of the cloud, due to the increased demand there might be created a “glut of
VMs”. VMs can be quickly delivered by cloud self-service portals. VMs can be transferred to previous
versions, can be paused and restarted. All of this can be done relatively easily. It is also possible to clone
them, and move between physical servers. Errors and vulnerabilities can be propagated without knowledge
about them. The difficult part might be maintaining record of the state of security at any point of time [1].
Recently, a member of Amazon Web Services uploaded and a pre-built image and this posed a threat on
whole Amazon community, since the image contained the publisher's secure shell (SSH) on it. This is
because the image could enable the publisher to log in to any machine that has the image. As a result, this
event made the use of pre-built machine images questionable, despite their handiness in saving time.
Conclusions
Virtualization and cloud computing help to eliminate traditional boundaries in networks. These new
technologies must support consumers with a widening scope of devices to access data in smart phones, tablet
computers, net books, notebooks, and traditional laptops. Cloud security architecture must adapt to these
shifting patterns, it must also support the infrastructure benefits of flexibility and cost savings.
This paper discussed threats in virtualization and cloud computing environments. Recommendations to
solve those issues are presented as well.
REFERENCES
1. Trend Micro. Security Threats to Evolving Data Centers. Retrieved from http://www.
trendmicro.com/cloud-content/us/pdfs/security-intelligence/reports/rpt_security-threats-to-datacenters.pdf
2.
K.Craig-Wood.
(2010).
IaaS
vs.
PaaS
vs.
SaaS
definition.
Retrieved
from
http://www.katescomment.com/iaas-paas-saas-definition/
3.
Target.
(2010).
Virtualization.
Retrieved
from
http://searchservervirtualization.
techtarget.com/definition/virtualization
4.
D.
L.
Ponemon.
(2010).
Security
of
Cloud
Computing
Users.Retrieveed
from
http://www.ca.com/us/~/media/files/industryresearch/security-cloud-computing-users_235659.aspx
5.
F.
Sabahi.
(2011).
Cloud
computing
security
threats
and
responses.
Retrieved
from
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6014715
6. J. W.Rittinghouse and J. F.Ransome. (2010 ). Cloud Computing: Taylor and Francis Group, LLC.
Retrieved from www.efgh. com/software/rijndael. htm
Иманбаева А.К., БисариновБ.Ж., Бисаринова А.Т.
Бұлттағы қауіпсіздік осалдықтар, қатерлержәне шешімдер
Түйіндеме. Ақпараттық технологиялар индустриясында бұлтты технологиялар өте қарқынды даму
үстіндегі секторлардың бірі болып табылады. Мұның себебі – есептеуіш процестерге кететін шығындардың
азаюуы және технологияның ыңғайлы болуы. Бұлттық технология көптеген ұйымдар арасында кең
тарағандықтан, бұлтқа қатысты маңызды мәселелерді қарастырған абзал. Бұлт провайдерлері көп кездесетін
мәселелердің бірі қауіпсіздік болып табылады. Бұл мақалада бұлттық технологиялар саласындағы қауіпсіздік
проблемалары қарастырылады және сол проблемалардың мүмкін шешімдері ұсынылады.
Достарыңызбен бөлісу: |