Modeling For Fault Tolerance in Cloud Computing Environment
Modeling For Fault Tolerance in Cloud Computing Environment
Modeling For Fault Tolerance in Cloud Computing Environment
Environment
Taskeen Zaidi Rampratap
Faculty of Computer Science & Engineering, Shri Ramswaroop Memorial University, Deva Road,
Lucknow, India
Abstract
Due to emergence of cloud computing, many traditional issues have been eliminated or
reduced. Cloud computing is on demand computing in which processing is done on remote
computer hence the chances of errors is more due to communication delay, latency or loss of
control over computing node. These are the important issue thats why cloud
computinginfrastructure should be fault tolerant as well as designed to schedule tasks must be
properly handled. The present paper deals with the understanding of fault tolerance techniques in
cloud environments and comparison with various models on various parameters have been
done. Fault tolerancetechnique is studied with the help of well known object-oriented language
Unified Modeling Language and state diagrams are designed and validated through the concepts
of Finite State Machine.
Keywords: UML, cloud computing, fault tolerance, UML, FSM, test cases,round robin
1. Introduction
Cloud computing is used for manipulating, configuring, and accessing the hardware and
software resources from remote location. It tenders online data storage, infrastructure,
and application. Cloud computing imparts by which we can access the applications as
utilities over the internet. It is used to create, configure, and customize the business
applications online. Fault tolerance is a major issue to guarantee the availability and
reliability of critical services as well as application processing and execution. In order to
depreciate failure impact on the system and application execution, failures should be
anticipated and proactively handled. Fault tolerance techniques are aiming to predict these
failures and take an appropriate action before failures actually occur. The present paper
overviews the existing fault tolerance techniques in cloud computing based on their
policies, tools used and research challenges.
Maliks and Fabrice [1]
have proposed the fault tolerance model for real time cloud
computing and analyzed on the basis of reliability of processing nodes. The systems
tolerate the faults and it is predicted that the reliability of virtual machines changes
frequently and it may be of adaptive nature. Pullum [2]
has described the concepts of
software fault tolerance techniques, fault removal, fault forecasting, checkpoints, and data
diverse software fault tolerant techniques. The fault tolerancetechniques with more
emphasis on reliability factor are well studied in [3]
. An innovative method for creating and
managing fault tolerance is proposed by Jhawar et al. [4]
. A generic fault
tolerancemechanism is for independent module which validates the users requirements.
Various factors like performance, fault tolerance, scalability of virtual management system
with three structure centralized, hierarchical and peer to peer are well studied by Kong et
al. [5]
. Guerraoui and Yabandeh [6] have predicted that resource graph and database are
necessary for service provider to ensure the behavior of fault tolerance mechanism. A
approach is developed by Jhawar et al. [7]
in which fault tolerance mechanism is evaluated
that uses virtualization to transparently increase the reliability and availability of cloud
infrastructure. Role of virtualization in cloud computing is well explained in Malhotra et
al. [8]
. Key reasons for virtualization and various standards of networking technologies and
concepts of software defined networking are described in [9]
. Lombardi and Pietro [10]
have
explained that security requirements are based on configuration management and
vulnerability assessment. In the present work, Unified Modeling Language (UML) is also
used to create a model. Let us describe some of the important review on UML which is a
general purpose modeling language that is used to model various kinds of the research
problem. It is widely accepted by the software professionals and developed by Object
Management Group (OMG) [11, 12]
. Modeling with real time characterization in UML,
approach and structure, viewpoint, UML viewpoint, concurrency modeling, performance
modeling, real time CORBA applications are well explained in [13]
. UML diagrams in
mathematical analyzable format and transformed into queuing network based on
intermediate textual representation has been represented in [14]
. Pllana and
Fahringer [15, 16]
have also used the UMLmodeling language for the performance oriented
distributed and parallel architecture applications.
In the present work, we have suggested different fault tolerance techniques and proposed
a fault tolerance technique using round robin scheduling scheme in which failures are
minimized on system through checkpoints when a fault occurs on system. Object oriented
language UML is used to design a model in terms of state transition diagram for the
systems processing and updating of local and global checkpoints and model is validated by
the various test cases.
2. Background
2.1. Fault Tolerance Techniques
There are two different types of fault tolerance techniques which are described below in
brief:
1. Checkpointing
If the task failed then it is restarted from recent checkpoint in spite of beginning. It is
efficient when implemented for large application.
2. Replications
It is used to maintain execution speed for different replicas of task run on different
resources until the replicated task not crashed. Various techniques like Haddop, HAP proxy
and Amazon EC2 are used to implement replicas.
3. Job Migration
It is used when failure occurs then job is migrated to a new machine.
4. Retry
It is simplest technique in which user resubmits task to same cloud resource.
5. Exception Handling
In this method user defines the specific action of task failures for workflows.
6. Task Resubmission
Failed task is submitted either on same machine or on another machine when it was
operated.
1. Self-Healing
If the instance of an application running or multiple virtual machines fails then it is
controlled automatically.
2. Software Rejuvenation
System starts with new state every time and also planned for periodic reboots.
3. Preemptive Migration
The application is constantly observed and analyzed and it depends upon feedback loop
control mechanism.
2.1.3. Hardware Fault Tolerance
Fault tolerant techniques are divided towards building computer that automatically recover
from random faults occurring in hardware components. By partitioning of comput er
system into pieces, the protective redundancy is used. Major approaches to recover
hardware failures are described below:
1. Fault Masking
It is structural redundancy technique that completely masks fault in redundant modules.
Triple Modular Redundancy (TMR) is used for fault masking in which circuitry triplicated
and voted. The individual voter failures are corrected by voting process.
2. Dynamic Recovery
Dynamic recovery is useful when only one copy of computation is running at a time and it
involves automated self-repair. In this approach, spiral mechanism is required to detect
failures in modules, switch out a faulty module, instigate actions to restore and continue
completion. Using this approach, computational delays the fault recovery, fault coverage is
lowered.
Global Checkpoint
Let us define global check point in which simulator assigns threads to the requested
processes and allocates sub clouds in round robin manner and global checkpoint will be
updated. It also tells the information about time of entry of job, number of processors
required, serial time, thread time, etc. The algorithm is given below:
1. Virtual Machine (VMs) submits their request to cloud service provider;
2. Cloud Service Provider(CSP) assigns threads to request and allocates minimum loaded
sub clouds in round robin way;
3. After allocation of subclouds global checkpoint will be updated periodically;
4. By reading checkpoint cloud service provider will check whether any subclouds has
failed or no failure occurs. If no failures occur then global checkpoint will be updated but if
failure occurs then job will be migrated to secondary node in round robin manner and then
global checkpoint will be updated.
The above steps are represented by the use of UML in the form of state transition model:
Local checkpoint
It tells information about server status, job status, server name and remaining time of
threads. The proposed algorithm is given below:
The total states are q 0, q1, q2, q3, and q4 with number of events as represented in above
table. Transition functions and equivalent grammar are generated from Figure 3, Figure 4,
which are given below:
q0aq1
q0aq2
By replacing non terminals on RHS of production we can get,
q0aaq2
(ii) Global checkpoint updating
q0aq1
q1bq3
q1bq4
By replacing non-terminals on RHS of production we can get,
q0abbq4
(iii) Local checkpoint updating
q0aq1
q1bq3
q4cq3
q3dq5
q4 dq5
By replacing non-terminals on RHS of production we can get,
q0abcddq5
(iv) GC updating after recovery
q0aq1
q1bq3
q4cq3
q3dq5
q4 dq5
By replacing non-terminals on RHS of production we can get,
q0abcddq5
(v) GC updating after recovery
q0aq1
q1bq3
q1bq4
q4cq3
By replacing non-terminals on RHS of production we can get,
q0abbcq3
q0aq1
q1bq2
q1bq3
q3cq4
q2dq4
By replacing non-terminals on RHS of production we can get,
q0abbcdq4
(ii) GC updating after recovery
q0aq1
q1bq2
q1bq3
q3cq4
By replacing non-terminals on RHS of production we can get,
q0abbcq4
5. Conclusion
Due to the dynamic nature of cloud environment system behavior will become unexpected
and results in faults or failures. Fault detection in clouds computing environment is one of
the biggest challenge now days.
The propose work emphasized towards the study of various fault tolerance techniques and
then detection of faults and it prevention by issuing checkpoints in round
robin manner, test cases are also designed for validation of proposed model. The faults
will be handled using round robintechnique with the updating of local and global
checkpoint.
Acknowledgement
In all humility and with much fervor, the authors are very thankful to Computer Science
and Engineering department, Shri Ramswaroop Memorial University, Lucknow to carry out
the above research work.
References
[1] Malik, Sheheryar and Fabrice Huet. Adaptive fault tolerance in real time cloud
computing services, 2011 IEEE world congress on. IEEE 2011.
[2] L.L. Pullum. Software fault tolerance and implementation Artech House, Boston,
London,UK 2001.
[3] S.Lakshmi Sudha. Fault tolerance in cloud computing IJESR International Journal of
Engineering Sciences Research, Vol 04, 2013.
[4] Jhawar, Ravi, Vincenzo Piuri & Marco Santambrogio, Fault tolerance management
in cloud computing. A system level perspective Systems Journal. IEEE 7.2(2013): 288-
297.
[5] X. Kong, J. Huang, C. Lin, P.D. Ungsunan, Performance, fault tolerance and scalability
analysis of virtual infrastructure management system, 2009 IEEE international
symposium on parallel and distributed processing with applications, Chengdu, China, Aug
9-12, 2009.
[6] R. Guerraoui and M. Yabandeh, Independent faults in the cloud in Proc. 4th international
workshop Large scale distributed system, Middleware, no 6.2010, pp12-17.
[7] Jhawar, Ravi, Vincenzo Piuri and Marco Santambrogio Fault Tolerance Management In
IAAS cloud. IEEE 2012.
[8] L.Malhotra, D. Agarwal & A.Jaiswal, Virtualization in cloud computing, JITSE, Volume 4,
Issue 2,ISSN: 2165-7866.
[9] R.Jain & S. Paul (2013) Network virtualization and software defined networking for cloud
computing: A survey IEEE 24-31.
[10] F.Lombardi, R.Di Pietro, Secure virtualization for cloud computing, Journal of network
and computer applications, Elsevier 2010.
[11] OMG, 2001, Unified Modeling Language Specification. Available online via
http://www.omg.org.
[12] OMG, 2002, OMG XML Metadata Interchange (XMI) Specification. Available online via
http://www.omg.org.
In article
[13] Object Management Group.UML Profile for Schedulability, Performance & Time
Specification. OMG Adopted Specification pt/02-03-02, Object Management Group,
March 2002.
In article
[15] S. Pllana, T. Fahringer, 2002, On Customizing the UML for Modeling Performance
Oriented Applications. In <<UML>>, Model Engineering Concepts and Tools, springer
Verlag., Dresden, Germany.
In article View Article
[16] S. Pllana, T. Fahringer, 2002, UML Based Modeling of Performance Oriented Parallel and
Distributed Applications, Winter Simulation Conference.