Analysis of Measures to Achieve Resilience during Virtual Machine

Interruptions in IaaS Cloud Service

Priya Vedhanayagam

, Subha S.

, Balamurugan Balusamy

, P. Vijayakumar

and Victor Chang

School of Information Technology and Engineering, VIT University, Vellore, Tamilnadu, India

University of College of Engineering Tindivanam, Melpakkam, Tamilnadu, India

International Business School Suzhou, Xi’an Jiaotong Liverpool University, Suzhou, 215123, China

Keywords: Cloud Computing, IaaS, Performance Evaluation, Queueing, Virtual machine, Resilience.

Abstract: In cloud computing era, the resilience issues faced by cloud computing services may be high. And therefore,

the best alternative to reckon with the effects on the Quality-of-Service is to preserve resilience of Cloud

computing service. To address this issue, an analytical model is proposed to study queueing system to

handle various virtual machine interruptions. The proposed model recommends a secondary virtual machine

to redeem the primary virtual machine during a probable halt. The work highlights the innovation employed

for analysing the measures to achieve resilience during virtual machine interruptions in IaaS cloud service,

the main objective of this research. The model is simulated using SHARPE and the results declare

guaranteed performance for the IaaS clients to achieve high availability of service as the response time

never deflate during VM interruptions.

1 INTRODUCTION

Cloud computing is the most advanced technology

in the realms of computing and its services are

attaining the stature of an intrinsic value that

governs one’s day to day life. Cloud services are

supported by a framework namely Internet Data

Center (IDC) (Armbrust, M.,et al., 2010). Nourished

by the broad accessibility of rapid web access, and

nurtured by the need to encourage clients to lessen

IT operational costs, the utilization of Cloud

computing has expanded widely in the past several

years. Cloud Computing depends on a service-

oriented architecture and renders three

classifications of services namely Infrastructure-as-

a-Service (IaaS), Platform-as-a-Service (PaaS), and

Software-as-a-Service (SaaS). The IaaS is concerned

with hardware, storage, servers, and networking

components over the Internet. The PaaS concentrates

on virtualized servers, operating systems, and other

hardware and software computing platforms. And

the SaaS delivers application software and other

services to host the application (Rimal, B. P., et al.,

2011). The entrepreneur’s chief aim is to achieve

more computing facilities and benefits with fewer

resources in different environments. The cloud

technology is a gift box loaded fully with benefits

like adaptability, disaster recovery, automatic

software upgrades, free capital-investment,

expanded collaboration, work from any-place,

document control, security, competitiveness and

ecologically friendly.

Resilience is transforming into an essential

service primitive for numerous cloud computing

applications. Metrics for resilience are greatly

associated with dependability metrics that are based

on availability, performance and survivability. In our

proposed model, resilience is defined as the

capability of a system to recover from various

virtual machine interruptions. To appraise resilience,

we exploit dependability attributes of systems such

as availability and performance (Javadi, B., et al.,

2013). This model makes a deep study of resilience

analysis of IaaS cloud (Ghosh, R.,et al., 2010)

considering various interruption states of virtual

machine. These interruptions result in downtime

which degrades the overall performance of the

system and violate the Quality-of-Service specified

in the Service Level Agreement and significantly

affect the availability of the system.

As per our knowledge, there is practically no

existing work that asserts these issues, as will be

seen in Section 2 below. To overcome these issues,

the proposed system models the cloud data centre

Vedhanayagam, P., S., S., Balusamy, B., Vijayakumar, P. and Chang, V.

Analysis of Measures to Achieve Resilience during Virtual Machine Interruptions in IaaS Cloud Service .

DOI: 10.5220/0006419904490460

In Proceedings of the 2nd International Conference on Internet of Things, Big Data and Security (IoTBDS 2017), pages 449-460

ISBN: 978-989-758-245-5

449

having

[]

//1

queueing system (Baba, Y, 1987)

as an internal queue that takes the batch of task

arrivals to a single virtual machine, on the

assumption that each task is serviced by a single

VM. During any of these interruptions, a secondary

virtual machine is quickly substituted to attend to the

scheduled task in the primary VM. This creates high

availability of resources without substantial

downtime.

The remainder of the paper is organized as

follows: Section 2 presents the related work in

performance analysis of cloud data centers; Section

3 depicts the evolution of our model and the details

of the analysis. Section 4 introduces an analytical

model considering different virtual machine

interruptions. Section 5 discusses the different

special conditions for analysing the parameters.

Section 6 lists out the numerical results obtained

from the analytical model and Section 7 summarizes

the results.

2 RELATED WORKS

Over the years, cloud computing has pulled in

extensive research attention, however just a small

portion of the work has been carried out to address

performance and resilience issues by analytical

models.

The primary commitment of the researchers

Khazaei, H., et al. (2011a) is to create performance

models for cloud computing centers. They proposed

a basic analytical model, using M/G/m queueing

system for cloud computing centers to examine the

most important aspects of present day cloud centers.

Based on this work, the authors had published the

book titled “Cloud Computing: Performance

Analysis” (Wang, L., et al. 2011) by extending their

work to consolidate the cloud’s focuses with a

hyper-exponential family. Later on, they included

presumptions of finite capacity for cloud center,

which made their model to be more like a genuine

cloud center (Khazaei, H., et al. 2011b).

Interestingly, Khazaei et al. (2012a) developed

M/G/m/m+r queueing system, a conceptual model to

manage the performance assessment of a cloud

center with the two-stage approximation technique.

Specifically, it permits precise modeling of cloud

centers with countless Physical Machines. They

extended their work (Khazaei et al. 2012b) to meet

the demands of challenging circumstances where

virtualization could be used to contribute a versatile

characterized set of computing resources that would

have high degree of virtualization. Khazaei, H., et al.

(2013a) had improved the proposed analytical model

to fasten critical perspectives such as pool

administration, power utilization, resource allocation

process and virtual machine organization of present

day cloud centers. Khazaei et al. (2013b) introduced

a performance model sensible for large IaaS clouds,

utilizing interactive stochastic models. Khazaei et al.

(2013c, 2013d) had furthermore presented an

interactive stochastic model which was realistic for

cloud computing centers with heterogeneous

demands and resources. Moreover, in particular, a

client’s task may ask for various sorts of VMs.

Bruneo, D., et al. (2010) exhibited a novel strategy

to interpret WS-BPEL forms into non-Markovian

stochastic Petri nets which, when permitted, would

systematically assess the importance of performance

indices primarily, and evaluate the performance of

web service at the initial design stage. The similar

work was tended by Bruneo et al. (2011) with an

objective of assessing various service parameters.

Bruneo et al. (2013a) assessed QoS oriented

performance analyses through the estimation of

steady-state measures and by inspecting the delays

presented in service endowment with an ultimate

aim of decreasing energy costs. Bruneo et al.

(2013b) offered an analytical model based on their

previous work that could undoubtedly actualize

resource allocation policies in a Green

Infrastructure-as-a-Service cloud. Bruneo et al.

(2013c) provided a dual solution for fluctuations in

the workload. One way was through the

conservation of reliability principle, and the other

was to optimize the Virtual Machine Monitor

software in case of an occurrence of workload

changes. Bruneo et al. (2014) proposed a stochastic

model to validate the performance of IaaS cloud

service.

Ghosh et al. (2010b) adopts a rapid and reasonable

technique for examining service excellence of huge

sized IaaS cloud after quantifying the effects of

variations in workload, fault load, system capacity

and developed submodels for failure, repair, and

migration of Physical Machines in Cloud using

scalable and highly reliable stochastic models.

Ghosh et al. (2010a) felt that resiliency

quantification would be a critical task and he

extends his previous work with an awareness to

manage non-homogeneous interacting sub-models.

The researcher‘s key inspiration for driving and

creating adaptable stochastic models for the cloud is

to help the service provider through what-if analyses

utilizing the execution model created from his

previous model (2010b). The Power-performance

trade-off analysis for IaaS Cloud (2011)

WICSPIT 2017 - Special Session on Innovative CyberSecurity and Privacy for Internet of Things: Strategies, Technologies, and

Implementations

450

demonstrates that natural gathering of Physical

Machines taking into account their power utilization

and response time conduct may not prompt craving

results and portray the cost breakdown and capacity

scheduling (Ghosh et al. ,2013; 2014a; 2014b).

Bacigalupo et al. (2011), Yang et al. (2013), Tian et

al. (2013), Khomonenko and Gindin (2014),

Vilaplana et al. (2014), Cao et al. (2014), Guo et al.

(2014), Cheng et al. (2015), Liu et al. (2015),

Sousa et al. (2015), Mei et al. (2015), Liu et al.

(2015), Xia et al. (2015a, 2015b), Liu, X et al.

(2014, 2015), Zhang et al. (2016), are the different

research works that deal with various queueing

models and provided diverse solutions for evaluating

performance measures.

As per our studies on the works published

previously, it is noted that this research work

effectively contributes to the analysis of measures

during the resilience of cloud services during

different VM interruptions. Our proposed work

applies Sumudu transform on M[x]/G/1 queueing

system to overcome certain deficiencies found in

other transforms like Fourier, Laplace, Mellin, etc.,

and also to make expressions simple, more intuitive

and applicable in different cases (Khalaf &

Belgacem, 2014).

3 SERVICE MODEL OF THE

PROPOSED SYSTEM

In an IaaS cloud service model, the IaaS providers

host client’s applications and deal with

responsibilities including system maintenance,

backup, and resiliency planning. IaaS platforms

strive to offer an on-demand scalable resource which

makes IaaS appropriate for inconsistent workloads.

Client’s workloads may be severely affected, when

an IaaS provider experiences downtime. Figure 1

shows the proposed IaaS cloud service model. An

IaaS provider’s responsibility is to provide service to

more and maximum clients with high availability. In

the IaaS cloud infrastructure, every physical

machine in the physical layer serves as a virtualized

environment through a Virtualization Server on top

of which one or more VMs can be instantiated. A

hypervisor decouples the VMs from the physical

host and allocates resources dynamically to each

VM as per the requirement to provide service for the

client’s task. Here, we assume that each task is

served by a single virtual instance, i.e., primary VM.

A Queueing system serves as an internal queue to

study the resilience of the system during different

VM interruptions.

Figure 2 shows the metrics Up Time and Down

Time of the system during different interruptions

like the vacation period of the primary VM, the

extended vacation period of the primary VM, the

repair period of primary VM and a delay time during

the repair period. During these interruptions, a

primary virtual machine can be supported by a

secondary virtual machine to achieve resilience

(VMware vLockstep, 2017). The secondary VM

works along with primary VM in perfect synchrony

and at the event of every primary VM interruption,

the interrupted task goes to the head of the queue.

An instance of secondary VM is created

automatically, and it handles the task from the

queue. We assume that once the primary VM

recovers from the interruption, the workload is

transferred back to the primary VM.

According to the problem design, from an

operational point of view, the primary VM can be

represented as a finite state machine categorized by

different operating states for different primary VM

interruptions as depicted in Figure 3. Once started,

the primary virtual machine is in the active working

state

with working event

wrk

e , it might enter into

any of the different interruption states. When the

primary VM goes on vacation, it enters state

with an event

e for the vacation period, and after

completing the vacation, it enters state

with an

event

and resumes the active state

. After

returning, the primary VM can go for an extension

of vacation, it enters state

with an event

e for

the extended vacation period, after completing the

extended vacation enters state

with an event

and resumes the active state

. In case of any

unexpected failure, the primary VM goes to

repairing, and enters state

with an event

under repair period, and after completing the

repairing process, enters state

with an event

and resumes active state

. The primary VM can

go to repair after some time, and enters state

with an event

e for a repair period after a

delay, and after completing the repairs it enters state

with an event e

and resumes active state

. In

the event of any interruption, the primary VM gets

automatically triggers the event

trig

to swap its state

Analysis of Measures to Achieve Resilience during Virtual Machine Interruptions in IaaS Cloud Service

451

Figure 1: IaaS cloud model supported by secondary VM with

[]

//1

internal queueing system.

Figure 2: Timing metrics during different primary VM interruptions.

WICSPIT 2017 - Special Session on Innovative CyberSecurity and Privacy for Internet of Things: Strategies, Technologies, and

Implementations

452

Figure 3: Operating states of primary VM during different interruptions.

to secondary VM service state

to handle the

workload. Once the primary VM completely

recovers, the secondary VM swaps its state

with an event

ret

e .

Tools of Queuing theory can be utilized

analytically to examine the behaviour of the system

described above and it resolves the most interesting

performance factors such as response time, task

blocking probability, probability of immediate

service, mean number of tasks, mean number of

customers, mean number of customers in the queue

and the mean waiting time in the system. In reality,

the service or server might face some accidental

failure or breakdown. In such contents, the service

provider can’t provide reliability and availability,

until the system recovers from the failure. The main

motive of our proposed work is to consider the

impact of resilience and great recovery options for

various virtual machine interruptions in an IaaS

cloud service. The IaaS clients should not

experience any downtime, so as to have a high

resilient system during different primary VM

interruptions.

4 QUEUEING MODEL OF THE

PROPOSED SYSTEM

In this work our focus was on

[]

//1

queueing

framework. Client’s tasks arrive in batches of size

to the “Primary VM” and given service in an FCFS

fashion

()

1, 2, 3...

= . When the Primary VM

becomes unavailable due to various interruptions, a

“Secondary VM” is assigned automatically to

provide continuous service to the clients in the same

manner. Primary VM provides service with

conditional probability

()

vdv

for a period of

time

()

,vv dv+ .

()

Δ is the probability of

Primary VM completing its idle period and

()

vdv

is the probability of Primary VM

completing its extended idle period. Breakdown

Analysis of Measures to Achieve Resilience during Virtual Machine Interruptions in IaaS Cloud Service

453

times of the primary virtual machine trail Poisson

distribution using mean breakdown rate,

. For

the period of primary VM failure, the task whose

service is hindered returns to the queue head,

however, it is immediately taken up for service by

the secondary VM and the repair procedure may

begin at any time.

()

Δ is the probability of the

Primary VM recovering after the repair and

()

is the probability of the Primary VM returning to the

active state after a delay time. The service rate of the

Secondary VM follows an exponential distribution

. When the primary VM recovers from the

interruptions, the task currently served by the

secondary VM that swaps over to the primary virtual

machine to begin the service. Every single stochastic

procedure required in the framework is autonomous.

5 PROPOSED ANALYTICAL

MODEL DEPICTING

DIFFERENT INTERRUPTIONS

OF THE PRIMARY VM

5.1 Service Time for a Virtual

Machine

() ()

()

() ()

jjkjk

Tv v Tv bT v

λμ ψ λ

−

∂

=− + + +

∂



(1)

() ()

()

Tv v Tv

λμ ψ

∂

=− + +

∂

(2)

5.1.1 Boundary Condition

() ( )

() ()

()

() ()

() () () ()

01 1

111

TmTvvdvnOvvvdv

Ev v v dv Ur v v dv b

jjj

μα

δωλ

∞∞

=− +− +



∞∞



+++

(3)

5.1.2 Probability Generating Function

()

() ()

{}

Su F q

qx F q m mA e xL

ηψ



−−





−−+ −



(4)

5.2 Primary Virtual Machine on

Idle Time

() () () () ()

Ov v v Ov v b Ov v Ov v

kjk

λα θ λ θ







∂

=− + + + +



−

∂

(5)

() ()

()

() ()

001

Ov v v Ov v Ov v

λα θ θ

∂

=− + + +

∂

(6)

WICSPIT 2017 - Special Session on Innovative CyberSecurity and Privacy for Internet of Things: Strategies, Technologies, and

Implementations

454

5.2.1 Boundary Condition

() () ()

Ov m T v v dv

∞



(7)

5.2.2 Probability Generating Function

()

() ()

{}

qmSu A

Ov x

qx F q m mA e xL



−−





−−+ −



(8)

5.3 Primary Virtual Machine on

an Extended Idle Time

() () () () ()

Ev v v Ev v b Ev v Ev v

kjk

λδ θ λ θ







∂

=− + + + +



−

∂

(9)

() ()

()

() ()

001

Ev v v Ev v Ev v

λδ θ θ

∂

=− + + +

∂

(10)

5.3.1 Boundary Condition

() () ()

Ev n Ov v v dv

∞



(11)

5.3.2 Probability Generating Function

()

() () ()

() ()

{}

** *

qmnSuF q A C

Ev x

qx F q m mA e xL

ηη



−−





−−+ −



(12)

5.4 Primary Virtual Machine is

under Repair

() () () () ()

Ur v v Ur v b Ur v Ur v

kjk

λω θ λ θ







∂

=− + + + +



−

∂

(13)

() ()

()

() ()

001

Ur v v Ur v Ur v

λω θ θ

∂

=− + + +

∂

(14)

5.4.1 Boundary Condition

() () ()

Ur Ds v v dv

∞



(15)

5.4.2 Probability Generating Function

()

() () ()

() ()

{}

** *

xSu F q D B

Ur x

qx F q m mA e xL

ψηη



−− −





−−+ −



(16)

Analysis of Measures to Achieve Resilience during Virtual Machine Interruptions in IaaS Cloud Service

455

5.5 Primary Virtual has

Delay in Recovery

() ()

()

() () ()

Ds v v Ds v b Ds v Ds v

kjk

λβ θ λ θ

∂

=− + + + +



−

∂

(17)

() ()

()

() ()

001

Ds v v Ds v Ds v

λβ θ θ

∂

=− + + +

∂

()

0Ds v

∂

(18)

5.5.1 Boundary Condition

() ()

Ds T v dv T

ψψ

∞

−−



()

00Ds =

(19)

5.5.2 Probability Generating Function

()

() ()

{}

xSu F q D

Ds x

qx F q m mA e xL

ψη



−− −





−−+ −



(20)

5.6 Overall System

( ) () () ( ) () () () ()

() ()

000

11S m T v v dv n Ov v v dv Ev v v dv

Ds v v dv

Ur v v dv

λμ αδ

∞∞∞

∞

=− +− + +





(21)

() () () () () ()

rr r r r r

Q x T x Ov x Ev x Ds x Ur x=+ + + +

(22)

Probability Generating Function

()

[]

{}

() ()

{}

***

Su F q x qmF q A e

qx F q m mA e xL

ηψ φ η

ηηψη

  

  

  







−− + + −

−−+ −

(23)

Substitute the subsequent values in the above

equations

()

qBax

λλ

=− + ,

()

Ba x

ηλλ θ

=− +− ,

() () ()

***

1LFqDB



=−



()

uBax

λλ

=− ,

()

1ennC

=− + ,

() ()

1 DB

ηη

=− ,

and

are

the Sumudu transforms used in the equations and the

Sumudu transform for any function is

() ( )

Sft futedt

∞

−







. During the applying

normalization condition, one finds S,

()

QS+=.

()

MQL Q x

, using this relation we can

find the steady state of Mean number of tasks served

in the virtual machine. Little’s law is applied to find

Mean waiting time of a task in the virtual machine,

MRT MQL

. Average number of task in the

queue is calculated as

AQL MQL

where

is the traffic intensity. Average waiting time of the

task in the queue can be calculated, using Little’s

Law

/ART AQL

(Little & Graves, 2008).

WICSPIT 2017 - Special Session on Innovative CyberSecurity and Privacy for Internet of Things: Strategies, Technologies, and

Implementations

456

6 NUMERICAL VALIDATIONS

The analytical model is simulated, using SHARPE

tool (Trivedi & Sahner, 2009). An extensive variety

of qualities were established for our model

parameters so that the model can state to an

expansive assortment of cloud service provider. We

assume that 1, 2, 4 or 8 Primary VMs are deployed

on a single Physical machine supported by parallel

Secondary VMs. The mean arrival rate of tasks to a

primary VM is

(we classify 500 to 1500 tasks

per hour). Mean Service time

(from

examination 30 minutes to 1 hour). Mean delay to

search a VM

(for current study 1 to 5 seconds).

Breakdown times of the virtual machine trail

Poisson distribution using mean breakdown rate

(few hours). According to the research work

to check the legitimacy of the results obtained, we

have studied the service time, idle times, delay

times, extended idle times and repair times and these

timings appear to be exponentially distributed. To

fulfil the stability conditions, all values have been

chosen subjectively.

Table 1: Performance measures for the proposed system

7, 5, 4, 2, 2, 0.5, 0.5.mn

αωλ

====== =

Table 2: Performance measures for the proposed system

7, 5, 4, 2, 2, 0.5, 0.5.mn

αωλ

====== =

Table 3: Performance measures for the proposed system

7, 9, 2, 2, 0.5, 0.5.mn

αλ

===== =

Table 4: Performance measures for the proposed system

7, 9, 2, 2, 0.5, 0.5.mn

αλ

===== =

Figure 4: Delay time Vs. Mean response time of the VM.

The impact of the delay times and extended vacation

times is depicted in Figure 4. Mean response time

decreases, even when there is an increase in the

delay rate, since secondary VM takes the

responsibility of primary VM when it is idle for a

long time.

Analysis of Measures to Achieve Resilience during Virtual Machine Interruptions in IaaS Cloud Service

457

Figure 5: Extended vacation time Vs. Mean number of

tasks in the queue.

The impact of the extended vacation times and Mean

number of tasks in the queue is depicted in Figure 5.

Mean number of tasks in the virtual machine

decreases, even when there is an increase in the

extended vacation times.

Figure 6: Secondary VM service Vs. Mean number of

tasks in the VM.

The impact of the secondary VM service times and

Mean number of tasks in the VM is depicted in

Figure 6. Mean number of tasks in the queue

decreases, even when there is an increase in the

secondary service times and the repair rate times.

Figure 7: Repair time Vs. Mean waiting time in queue.

The impact of the repair times and Mean waiting

time in the queue is depicted in Figure 7. Mean

waiting time in the queue decreases, even when

there is an increase in the repair times and the

secondary service rate times.

7 CONCLUSION

In this paper, we studied the resilience for IaaS

cloud and proposed an analytical model which

probes deep into our modern data centre to bring

new novelties. We are quite hopeful that this is an

innovative and honest attempt in analysing the

measures for the resiliency of IaaS cloud by

considering

[]

//1

queueing system and

presents exceptional performance measures during

distinct interruptions of primary virtual machine

supported by a secondary virtual machine by

utilizing the advantage of Sumudu transform. In

future, we plan to extend our work in

[]

//1

queueing system, as the task refuses to join the

queue, i.e., balking; and the tasks leave the queue

after entering, i.e., reneging during different virtual

machine interruptions.

REFERENCES

Armbrust, M., Fox, A., Griffith, R., Joseph, A. D., Katz,

R., Konwinski, A., & Zaharia, M. (2010). A view of

cloud computing. Communications of the

ACM, 53(4), 50-58.

WICSPIT 2017 - Special Session on Innovative CyberSecurity and Privacy for Internet of Things: Strategies, Technologies, and

Implementations

458

Baba, Y. (1987). On the Mx/G/1 queue with and without

vacation time under non-preemptive last-come first-

served discipline. J. of Opns. Res. Society of

Japan, 30, 150-159.

Bacigalupo, D. A., Van Hemert, J., Chen, X., Usmani, A.,

Chester, A. P., He, L., ... & Jarvis, S. A. (2011).

Managing dynamic enterprise and urgent workloads

on clouds using layered queuing and historical

performance models. Simulation Modelling Practice

and Theory, 19(6), 1479-1495.

Bruneo, D. (2014). A stochastic model to investigate data

center performance and qos in iaas cloud computing

systems. IEEE Transactions on Parallel and

Distributed Systems, 25(3), 560-569.

Bruneo, D., Distefano, S., Longo, F., & Scarpa, M. (2010,

April). Qos assessment of ws-bpel processes through

non-markovian stochastic petri nets. In Parallel &

Distributed Processing (IPDPS), 2010 IEEE

International Symposium on (pp. 1-12). IEEE.

Bruneo, D., Distefano, S., Longo, F., & Scarpa, M.

(2013c, October). Stochastic evaluation of QoS in

service-based systems. IEEE Transactions on Parallel

and Distributed Systems, 24(10), 2090-2099.

Bruneo, D., Distefano, S., Longo, F., Puliafito, A., &

Scarpa, M. (2013a, June). Workload-based software

rejuvenation in cloud systems. IEEE Transactions on

Computers, 62(6), 1072-1085.

Bruneo, D., Lhoas, A., Longo, F., & Puliafito, A. (2013b,

September). Analytical evaluation of resource

allocation policies in green iaas clouds. InCloud and

Green Computing (CGC), 2013 Third International

Conference on(pp. 84-91). IEEE.

Bruneo, D., Longo, F., & Puliafito, A. (2011, June).

Evaluating energy consumption in a cloud

infrastructure. In World of Wireless, Mobile and

Multimedia Networks (WoWMoM), 2011 IEEE

International Symposium on a(pp. 1-6). IEEE.

Cao, J., Li, K., & Stojmenovic, I. (2014). Optimal power

allocation and load distribution for multiple

heterogeneous multicore server processors across

clouds and data centers. IEEE Transactions on

Computers, 63(1), 45-58.

Cheng, C., Li, J., & Wang, Y. (2015). An energy-saving

task scheduling strategy based on vacation queuing

theory in cloud computing. Tsinghua Science and

Technology, 20(1), 28-39.

Ghosh, R., Longo, F., Frattini, F., Russo, S., & Trivedi, K.

S. (2014a, March). Scalable analytics for iaas cloud

availability. IEEE Transactions on Cloud

Computing, 2(1), 57-70.

Ghosh, R., Longo, F., Naik, V. K., & Trivedi, K. S.

(2010a, October). Quantifying resiliency of iaas cloud.

In Reliable Distributed Systems, 2010 29th IEEE

Symposium on (pp. 343-347). IEEE.

Ghosh, R., Longo, F., Naik, V. K., & Trivedi, K. S.

(2013). Modeling and performance analysis of large

scale iaas clouds. Future Generation Computer

Systems, 29(5), 1216-1234.

Ghosh, R., Longo, F., Xia, R., Naik, V. K., & Trivedi, K.

S. (2014b, December). Stochastic model driven

capacity planning for an infrastructure-as-a-service

cloud. IEEE Transactions on Services

Computing, 7(4), 667-680.

Ghosh, R., Naik, V. K., & Trivedi, K. S. (2011, June).

Power-performance trade-offs in IaaS cloud: A

scalable analytic approach. In 2011 IEEE/IFIP 41st

International Conference on Dependable Systems and

Networks Workshops (DSN-W) (pp. 152-157). IEEE.

Ghosh, R., Trivedi, K. S., Naik, V. K., & Kim, D. S.

(2010b, December). End-to-end performability

analysis for infrastructure-as-a-service cloud: An

interacting stochastic models approach. In Dependable

Computing (PRDC), 2010 IEEE 16th Pacific Rim

International Symposium on (pp. 125-132). IEEE.

Guo, L., Yan, T., Zhao, S., & Jiang, C. (2014). Dynamic

performance optimization for cloud computing using

m/m/m queueing system. Journal of Applied

Mathematics, 2014.

Javadi, B., Thulasiraman, P., & Buyya, R. (2013).

Enhancing performance of failure-prone clusters by

adaptive provisioning of cloud resources. The Journal

of Supercomputing, 63(2), 467-489.

Khalaf, R. F., & Belgacem, F. B. M. (2014, December).

Extraction of the Laplace, Fourier, and Mellin

transforms from the Sumudu transform. In AIP Conf.

Proc (Vol. 1637, No. 142, pp. 6-1432).

Khazaei, H., Jelena, M., & Vojislav, B. M. (2013d).

Performance evaluation of cloud data centers with

batch task arrivals. Communication Infrastructures for

Cloud Computing, 199-223.

Khazaei, H., Misic, J., & Misic, V. B. (2011a, June).

Modelling of cloud computing centers using m/g/m

queues. In 2011 31st International Conference on

Distributed Computing Systems Workshops (pp. 87-

92). IEEE.

Khazaei, H., Misic, J., & Misic, V. B. (2011b, December).

Performance analysis of cloud centers under burst

arrivals and total rejection policy. InGlobal

Telecommunications Conference (GLOBECOM

2011), 2011 IEEE (pp. 1-6). IEEE.

Khazaei, H., Misic, J., & Misic, V. B. (2012a).

Performance analysis of cloud computing centers

using m/g/m/m+ r queuing systems. IEEE

Transactions on parallel and distributed

systems, 23(5), 936-943.

Khazaei, H., Misic, J., & Misic, V. B. (2013b, November).

A fine-grained performance model of cloud computing

centers. IEEE Transactions on parallel and distributed

systems, 24(11), 2138-2147.

Khazaei, H., Misic, J., & Misic, V. B. (2013c, December).

Performance of cloud centers with high degree of

virtualization under batch task arrivals. IEEE

Transactions on Parallel and Distributed

Systems, 24(12), 2429-2438.

Khazaei, H., Mišić, J., Mišić, V. B., & Mohammadi, N. B.

(2012b, December). Availability analysis of cloud

computing centers. In Global Communications

Conference (GLOBECOM), 2012 IEEE (pp. 1957-

1962). IEEE.

Analysis of Measures to Achieve Resilience during Virtual Machine Interruptions in IaaS Cloud Service

459

Khazaei, H., Mišić, J., Mišić, V. B., & Rashwand, S.

(2013a, May). Analysis of a pool management scheme

for cloud computing centers. IEEE Transactions on

parallel and distributed systems, 24(5), 849-861.

Khomonenko, A. D., & Gindin, S. I. (2014, October).

Stochastic models for cloud computing performance

evaluation. In Proceedings of the 10th Central and

Eastern European Software Engineering Conference in

Russia (p. 20). ACM.

Little, J. D., & Graves, S. C. (2008). Little's law.

In Building intuition (pp. 81-100). Springer US.

Liu, M., Dou, W., Yu, S., & Zhang, Z. (2015). A

decentralized cloud firewall framework with resources

provisioning cost optimization. IEEE Transactions on

Parallel and Distributed Systems, 26(3), 621-631.

Liu, X., Li, S., & Tong, W. (2015). A queuing model

considering resources sharing for cloud service

performance. The Journal of Supercomputing,71(11),

4042-4055.

Liu, X., Tong, W., Zhi, X., Zhiren, F., & Wenzhao, L.

(2014). Performance analysis of cloud computing

services considering resources sharing among virtual

machines. The Journal of Supercomputing, 69(1), 357-

374.

Liu, X., Zha, Y., Yin, Q., Peng, Y., & Qin, L. (2015).

Scheduling parallel jobs with tentative runs and

consolidation in the cloud. Journal of Systems and

Software, 104, 141-151.

Rimal, B. P., Jukan, A., Katsaros, D., & Goeleven, Y.

(2011). Architectural requirements for cloud

computing systems: an enterprise cloud approach.

Journal of Grid Computing, 9(1), 3-26.

Sousa, E., Lins, F., Tavares, E., Cunha, P., & Maciel, P.

(2015). A modeling approach for cloud infrastructure

planning considering dependability and cost

requirements. IEEE Transactions on Systems, Man,

and Cybernetics: Systems, 45(4), 549-558.

Tian, Y., Lin, C., Chen, Z., Wan, J., & Peng, X. (2013).

Performance evaluation and dynamic optimization of

speed scaling on web servers in cloud

computing. Tsinghua Science and Technology, 18(3),

298-307.

Trivedi, K. S., & Sahner, R. (2009). SHARPE at the age of

twenty two. ACM SIGMETRICS Performance

Evaluation Review, 36(4), 52-57.

Vilaplana, J., Solsona, F., Teixidó, I., Mateo, J., Abella, F.,

& Rius, J. (2014). A queuing theory model for cloud

computing. The Journal of Supercomputing, 69(1),

492-507.

VMware vLockstep, VMware Inc. vSphere ESX and ESXi

Info Center. http://www.vmware. com/products/esxi-

and-esx/overview.html, July 2017.

Wang, L., Ranjan, R., Chen, J., & Benatallah, B. (Eds.).

(2011). Cloud computing: methodology, systems, and

applications. CRC Press.

Xia, Y., Zhou, M., Luo, X., Pang, S., & Zhu, Q. (2015a,

January). A stochastic approach to analysis of energy-

aware DVS-enabled cloud datacenters. IEEE

Transactions on Systems, Man, and Cybernetics:

Systems, 45(1), 73-83.

Xia, Y., Zhou, M., Luo, X., Pang, S., & Zhu, Q. (2015b,

April). Stochastic modeling and performance analysis

of migration-enabled and error-prone clouds. IEEE

Transactions on Industrial Informatics, 11(2), 495-

504.

Yang, B., Tan, F., & Dai, Y. S. (2013). Performance

evaluation of cloud service considering fault

recovery. The Journal of Supercomputing, 65(1), 426-

444.

Zhang, S., Qian, Z., Luo, Z., Wu, J., & Lu, S. (2016).

Burstiness-Aware Resource Reservation for Server

Consolidation in Computing Clouds. IEEE

Transactions on Parallel and Distributed

Systems, 27(4), 964-977.

WICSPIT 2017 - Special Session on Innovative CyberSecurity and Privacy for Internet of Things: Strategies, Technologies, and

Implementations

460