Proactive Scheduling in Cloud Computing

Bulletin of Electrical Engineering and Informatics
ISSN: 2302-9285
Vol. 6, No. 2, June 2017, pp. 174~180, DOI: 10.11591/eei.v6i2.649  174
Received March 7, 2017; Revised May 8, 2017; Accepted May 22, 2017
Proactive Scheduling in Cloud Computing
Ripandeep Kaur*
1
, Gurjot Kaur
2
Department of CSE Chandigarh University, Gharuan, Punjab, India
*Corresponding author, e-mail: ripansandhu180@gmail.com
1
, randhawa789 @gmail.com
2
Abstract
Autonomic fault aware scheduling is a feature quite important for cloud computing and it is
related to adoption of workload variation. In this context, this paper proposes an fault aware pattern
matching autonomic scheduling for cloud computing based on autonomic computing concepts. In order to
validate the proposed solution, we performed two experiments one with traditional approach and other
other with pattern recognition fault aware approach. The results show the effectiveness of the scheme.
Keywords: fault tolerance, scheduling, performance metrics, cloud computing, QoS
1. Introduction
Cloud computing is a recent advancement wherein IT infrastructure and applications
are provided as „services‟ to end-users under a usage-based payment model. It can leverage
virtualized services even on the fly based on requirements (workload patterns and QoS) varying
with time [1]. According to NIST definition: “Cloud computing (CC) is a model for enabling
convenient, on-demand network access to a shared pool of configurable computing resources
(e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and
released with minimal management effort or service provider (SP) interaction [2].” Cloud service
users demand for their services end-to-end QoS assurance, high levels of service reliability, and
continued availability to their SPs. Now a days, IT enterprise is adopting cloud computing in
order to reduce the total cost involved and also improve the QoS delivered to the customers.
There are no standard metrics or a standard way to ensure QoS to the customers.
There are several models or algorithms that are proposed to ensure QoS to the users and
proper management of workloads to provide QoS and performance. So in CC, there are various
important research issues which need to be focused for its efficient performance is fault
tolerance and scheduling [3].
There has been various types of scheduling algorithm exist in cloud computing system.
Most of them can be applied in the cloud environment with suitable verifications. The main
advantage of job scheduling algorithm is to achieve a high performance computing and the best
system throughput. During scheduling, existing algorithms are not fully capable to evaluate the
fault and take decisions accordingly. Multiple reasons exist for low performance in scheduling
algorithms. Majority of literature focussed on the work to decrease response time in order to
provide scheduling in the cloud environments with Quality of Service (QoS). Therefore
performance is definitely one of the major concerns in using existing scheduling algorithm, but
improving performance with enhancing the fault tolerance of the cloud system is one of the
major research area which is not been explored very well [4-5]. To provide guaranteed Quality
of Service (QoS) to users, it is necessary that jobs should be efficiently mapped to
given resources.
Service Level Agreement (SLA) is the major parameter i.e. considered for assuring QoS
and it is responsibility of SPs whether at infrastructure, platform or software level- provide
quality guarantees usually in terms of availability and performance to their customers in the form
of SLAs. So it should be fault-tolerant, and recovery time should be minimal to avoid SLA
violation. The replica should be maintained near the customer‟s location to reduce the recovery
time after any failure or disaster. So SLA should include the availability, response time, and
degree of support [6].

Bulletin of EEI ISSN: 2302-9285 
Proactive Scheduling in Cloud Computing (Ripandeep Kaur)
175
This research paper proposes a service ranking algorithm in a CC on the basis of
detailed performance monitoring and historical analysis and based on their contribution, a
weight age is assign to all service quality factors or performance metrics and as a final point
aggregated to compute ranking score (R) of a service by developed formula. This new model is
used for VM allocation, re-allocation and placement with consideration of best/high ranked
virtual machine/datacenter available. Workload requested by the users under pre-analyzing the
job requests and the resource status of the data center considering various parameters like
Reliability, Reputation, Network Latency, Processing time, Availability etc.
We further summarize our objectives as under:
a. To develop a system that pre assume the time consumption of workload to identify the high
ranked VMs/DCs.
b. To establish the effectiveness of the system in correspondence to Service Level
Agreements (SLAs) Violations. We will be evaluating the impact of faults on scheduling and
improving scheduling by minimization of SLA violations.
c. To develop a QoS system for users in terms of response time i.e. time taken from the Cloud
to respond user‟s request.
2. Literature Review
Stelios Sidiroglou [7] et al presented an ASSURE, a new self-healing software (s/w)
based approach that presents rescue points (RP) for detecting/tolerating/recovering from s/w
failures in server apps while preserving system availability and integrity. Using fuzzing, they
identify rescue point and implemented by checkpoint/restart technique. When fault detects
initially, it uses an application replica to find out what RPs can be employed for recover
execution of future programs. This approach implemented on various applications of server like
proxy servers, domain name, database and web. The main goal of this approach is to healing
s/w services automatically from s/w failures that were previously unidentified or not known.
Hai Jin [8] et al introduce a SHelp, a novel self-healing s/w based approach which is
considering extension of original approach ASSURE that applies error virtualization and
weighted RP (WRP) methods which helps server applications to avoid faulty path. It can survive
s/w failures and to make sure high availability of service in CC environment. SHelp presents two
approaches. First, WRPs for recover from faults that are complicated to handle for ASSURE.
Second, to adopt two-level RP database which helps to share information related to faults with
applications that are helpful for further faults recovery.
Sheheryar Malik [9] et al proposed an AFTRC (Adaptive FT in Real time CC) model. By
computing reliability(R) of every VM, a system tolerates faults. After every cycle, reliability of
every VM is changed because of its adaptive behavior. A main goal of this model is to assign R
weights to every VM and removing/adding a VM, if it is not performing efficient in real time
environment. AFTRC also provides backward/forward recovery in case if any VM doesn‟t
achieve minimum reliability level and it also uses replication technique to achieve FT.
Dilbag Singh [10] et al proposed a smart failover approach for offering high availability
to the cloud‟s customers by using new algorithm namely; integrated checkpointing with load
balancing (ICWLB) and to reduce overheads of checkpointing by using multilevel checkpoint. A
proposed strategies used two different algorithms namely; global and local checkpointing
algorithms. This approach has been made performance comparison of various metrics like
Maximum/Minimum Execution time (Max/Min ET), Maximum/Minimum Waiting time (Max/Min
WT) with some existing methods and also shows a proposed strategy gives better results than
other strategies.
Pranesh Das [11] et al proposed a smart failover approach namely; Virtualization FT
(VFT) to attain the FT by using redundancy or replication technique. They presented a
virtualization technique where the Load balancer (LB) distributing loads to those nodes whose
related computing nodes have excellent performance history which further measure by using
Success rate of those computing nodes. This model helps to decrease timing of services and to
improve the availability by decision maker and cloud manager modules.
Deepak Poola [12] et al proposed a scheduling algorithm to schedule workflow tasks or
jobs on CC resources with the help of spot instances (SI) and on-demand instances (ODI)
pricing models and also to reduce execution cost in case of tasks deadline. A proposed
algorithm is used bidding method to decrease cost and bids according to the requirement of

 ISSN: 2302-9285
Bulletin of EEI Vol. 6, No. 2, June 2017 : 174 – 180
176
workflow. They also tolerate faults against early extinction of SI and robust against CC
instances variations in performance. This work saves cost upto 14% by using
checkpointing technique.
Mohammed Amoon [13] proposed an economy based FT framework to maintain
monetary profit by providing dynamic number of replicas and to tolerate faults for avoiding
failures. A main work presented by two algorithms namely VMC (VM Classification) and FTSS
(FT Strategy Selection). VMC classifies cloud VMs by using available information of usage
service time and probability of failures VMs and to select most valuable VM that are profitable
for cloud. FTSS is basically used for selecting suitable FT approach for selected virtual machine
that depends on requirements of customers like time deadline and cost of cloud applications.
This framework used various FT approaches like Proactive and Reactive FT and provide hybrid
FT. In Reactive, it uses various strategies like checkpointing, replication and further used
parallel and multiversion mechanisms of replication strategy.
Anju Bala [14] et al proposed an Autonomic FT (AFT) scheduling approach to assist the
execution of parallel tasks in cloud computing applications like scientific workflows (SW). Cloud
Service providers involve well-organized scheduling fault tolerant (FT) and Hybrid heuristics
(HH) techniques. HH merges the various features of FCFS, Min-Min and Max-Child heuristic. In
FT technique, due to over-consumption of resources if task failure happens then VM migration
(VMM) automatically migrates the VM. AFT approach significantly reducing make-span,
standard deviation and total mean execution time and improve performance of SW.
Punit Gupta [15] et al proposed a FLHB Scheduling algorithm for cloud IaaS. It provides
higher quality of services to the customer with least cost and also considers various datacenters
quality of service parameters like System load (MIPS), Network load, initialization time and Fault
Rate for improving the performance and quality of services to the customer in cloud
infrastructure environment.
3. System Model
In this section our proposed system model which explains Fault Aware Scheduling
Technique (FAST), as shown in Figure 1. where workload generator is responsible for creating
workloads. It is similar to the users who are requesting for VMs. These users defined a set of
quality parameters which needs to be met by any system.
Figure 1. Proposed system model
Therefore a model is required which involved the following steps:
a. A monitoring application collects the following values and after retrieving monitored values
fuzzy prediction process is initiated which sets the min and max performance of VM
components, i.e for each request to process there is requirement for CPU which ranges
[CPUmin∼CPUmax], similarly for Memory [MEMmin∼MEMmax] and for Bandwidth
[BWmin∼BWmax], after obtaining the degree of truthness for each component of VM
(fuzzification) these values get put into LOG.

177
b. Clustering of VMs: Each VM, with a common set of configuration is put into a common
cluster.
c. A commonly used VM allocation policy (Round Robin algorithm) is used to allocate the
incoming request to these clusters.
d. A constant tracking of SLA violations is done and in the event of any positive sign, a pattern
of VM working is obtained by comparing the current value with the LOG.
e. The pattern algorithm which is based on density-based spatial clustering [10] which
identifies the distribution of data in the current cluster and generates a trigger in the case of
any change required. Hence the first SLA violation is acting as the threshold value and is
represented by ε.
f. After getting the first SLA, the process of inputting is started and obtained results are
refuzzified. After this the current performance is logged in. Now this new cluster is used for
scheduling new job requests, which is done by identified faulty VMs who are not meeting
the requirements of users.
g. After identifying these which helps load balancer to take decision by redirecting the
incoming request to the VM who are working up to their capacity and very minimal requests
is inflow towards faulty VMs.
Algorithm 1: Clustering
For every VMi
Resi=Get_monitored_result(VM);//[CPUmin∼CPUmax], [MEMmin∼/ //MEMmax], [BWmin
∼ BWmax],
LOG(Resi);
Set(min,max)I // Fuzzification
Get_result=Match_Cluster_to_VMi (Cluster_Name, VMi)
If (Get_result=TRUE)
Set_ VMi(Cluster_Name)
endIf
Algorithm 2: Tracking Faults
Set_threshold=first_SLA_violation (VMi)
If (SLA_violation==TRUE)
Set(min,max) // ReFuzzification
Get_result= Match_Cluster_to_VMi (Cluster_Name, VMi)
If (Get_result=TRUE)
Set_ VMi (Cluster_Name)
endIf
4. Experimental Set up and Results
Following are the simulation parameters:
Number of Datacenters: 4
Number of Host/DC: 1
Number of VM/Host: 4
a. Experiment No. 1
In this experiment, a fault is created by lowering the CPU capacity, which directly
lowered the CPU consumption. In Figure 2. we can observe the CPU consumption where VM 1
is offering lowered capacity.

 ISSN: 2302-9285
178
Figure 2. Average CPU consumption
b. Experiment No. 2
In this experiment we analyze the response time of commonly used round robin
algorithm. We can observe that with the introduction of faults the response time is increased
drastically. Therefore the average response time i.e. 7.9 ms for the all the requests to process
shown in Figure 3 which is very high.
Figure 3. Response time
c. Experiment No. 3
In Figure 4, SLA violation is observed for the round robin algorithm which is very high
for VM1.
Figure 4. SLA violation for VM allocation policy
d. Experiment No. 4
In this experiment, again a fault is created to observe the behavior of proposed
algorithm (FAST). An average CPU usage for 2 VMs is constant but for third VM it is fluctuating,
hence underperforming CPU observed. Hence a optimal requests are being allotted, because
the workload was better distributed among the better performing virtual machines and also the

179
response time of requests initially had the lowest values, and most requests were allocated to
the VM 2, 3, and 4. The increase in workload leads to more allocations to only these VMs. Also
the observed Response time in this case is 4.46 ms.
This directly corresponds to the performance of SLA violations as we can observe in
Figure 7. all the VMs are performing equally. This is done by moving under performing VM to its
right cluster.
Figure 5. VM performance indicator
Figure 6. Response time for FAST
Figure 7. SLA violations
Both experiments used the same workload and resource allocation strategy. However,
the thresholds were different because of differnt SLA violations.

 ISSN: 2302-9285
180
5. Conclusion Future Work
Fault aware cloud computing environments to support the elastic provisioning has
proved to be very beneficial. Experiments conducted for validating the architecture clearly depict
that autonomic computing and cloud computing can be used together with various technologies
and different providers. The future work involves different criteria that should be used for rules
design (e.g., average response time of requests or latency). Furthermore, the use of other levels
of control loops may improve the architecture‟s effectiveness, focusing on better performance.
References
[1] Q Zhang, L Cheng, R Boutaba. Cloud Computing: State-of-the-art and Research Challenges. In Cloud
Computing. The Brazilian Computer Society Conference on Springer. 2010: 7-18.
[2] G Shroff. Enterprise Cloud Computing Technology, Architecture, Applications, Cambridge South Asian
ed., 2011: 51-60. ISBN: 978-1-107-64889-0.
[3] A Bahga, V Madisetti. Cloud Computing A hands-on Approach. Universities Press, 1st ed., 2014: 117-
120. ISBN: 978-81-7371-923-3.
[4] A Ganesh, Dr M Sandhya, Dr S Shankar. A Study on Fault Tolerance Methods in Cloud Computing. In
International Advance Computing Conference (IACC), 2014 IEEE Conference on 2014: 844-849.
[5] V Kumar, S Sharma. A Comparative Review on Fault Tolerance Methods and Models in Cloud
Computing. In International Research Journal of Engineering and Technology (IRJET). Nov 2015;
2(8): 1-7.
[6] K Chandrasekaran, Essentials of Cloud Computing, CRC Press, 3
rd
Ed., 2015: 49-60. ISBN: 978-1-
4822-0544-2.
[7] S Sidiroglou, O Laadan, C Perez, N Viennot, J Nieh, AD Keromytis. Assure: Automatic Software Self-
healing Using Rescue Points. In ACM Sigplan Notices. 2009; 44(3): 37-48.
[8] G Chen, H Jin, D Zou, BB Zhou, W Qiang, G Hu. SHelp: Automatic Self-healing for Multiple
Application Instances in a Virtual Machine Environment. In Cluster Computing (CLUSTER), 2010
IEEE International Conference on IEEE. 2010: 97-106.
[9] S Malik, F Huet. Adaptive Fault Tolerance in Real Time Cloud Computing. In Services (SERVICES),
2011 IEEE World Congress on IEEE. 2011; 280-287.
[10] D Singh, J Singh, A Chhabra. High Availability of Clouds: Failover Strategies for Cloud Computing
using Integrated Checkpointing Algorithms. IEEE International Conference on Communication
Systems and Network Technologies, 2012.
[11] P Das, PM Khilar. VFT: A Virtualization and Fault Tolerance Approach for Cloud Computing. In
Information & Communication Technologies (ICT), 2013 IEEE Conference on, 2013: 473-478.
[12] D Poola, K Ramamohanarao, R Buyya. Fault-Tolerant Workflow Scheduling Using Spot Instances on
Clouds. 14th International Conference on Computational Science (ICCS), Elsevier. 2014: 523–533.
[13] M Amoon. A Framework for Providing a Hybrid Fault Tolerance in Cloud Computing. Science and
Information Conference, London, UK, July 2015: 28-30.
[14] A Bala, I Chana. Autonomic Fault Tolerant Scheduling Approach for Scientific Workflows in Cloud
Computing. Concurrent Engineering: Research and Applications, SAGE. 2015: 1-13.
[15] P Gupta, SP Ghrera. Load and Fault Aware Honey Bee Scheduling Algorithm for Cloud Infrastructure.
Proc. of the 3rd Int. Conf. on Front. of Intell. Comput. (FICTA) 2014, Springer. 2015: 135-143.
[16] Cloud Service Measurement Index Consortium (CSMIC), SMI framework. [Last accessed:] 2/15, 2015,
[Online]. Available: http://beta-www.cloudcommons.com/servicemeasurementindex.
[17] SK Garg, S Versteegb, R Buyya. A Framework for Ranking of Cloud Computing Services. Future
Generation Computer Systems, Elsevier. 2013; 29(4): 1012–1023.
[18] K Elissa. Title of paper if known. unpublished.
[19] R Nicole.Title of paper with only first word capitalized. J. Name Stand. Abbrev. in press. University
Science, 1989.

Proactive Scheduling in Cloud Computing

More Related Content

What's hot (19)

Similar to Proactive Scheduling in Cloud Computing (20)

More from journalBEEI (20)

Recently uploaded (20)

Proactive Scheduling in Cloud Computing