A Resource Aware Load Balancing Model in Cloud Computing with Multi-Objective Scheduling

Transcription

1 A Resource Aware Load Balancing Model in Cloud Computing with Multi-Objective Scheduling Kavita Research Scholar Computer Science and Engineering SVIET Banur, Punjab, India Vikas Zandu Assistant Professor Computer Science and Engineering SVIET Banur, Punjab, India Abstract: Cloud computing is a service based, on-demand, pay per use model consisting of an inter-connected and virtualizes resources delivered over internet. In cloud computing, usually there are number of jobs that need to be executed with the available resources to achieve optimal performance, least possible total time for completion, shortest response time, and efficient utilization of resources etc. Hence, job scheduling is the most important concern that aims to ensure that use s requirement are properly and correctly satisfied by cloud infrastructure. Most of the task scheduling algorithm developed in cloud computing target single criteria which fails to provide efficient resource utilization. Soto enhances the system performance and increase resource utilization it is must to consider multiple criteria. This paper proposes a multiobjective scheduling algorithm that considers wide variety of attributes in cloud environment. The paper aims to improve the performance by reducing the load of a virtual machine (VM) by using Load Balancing Method. Finally, it optimizes the resource utilization by using Resource Aware Scheduling Algorithm involving combination of Min- max and Max-Min strategy. Keywords: VM, QoS, Non-dominated sorting, Multi-objective optimization, RASA algorithm I. INTRODUCTION A rapid change has been observed in the dispersed computing with time. Initially beginning with desktop computing, then grid computing and further moving into cloud computing has completely changed the whole computing definition. Since last few years, cloud computing has been widely adopted into business institutions, research institutions, industries and in academics. The rapid development of cloud computing has brought a bright prospect and more economic benefits to the commercial industries. Cloud Computing is intended to enable the computing across largely and diverse resources and dynamic stability. Cloud computing is a computing method that uses internet to provide the user with infrastructure, software and other IT services as per the user s demand with minimum expenditure in minimum time effort. In this computing model, users use the needed services without being bothered to know how these services are delivered or where services are hosted. With the increasing number of users in cloud computing, the volume of data is also expanding. The resource demands for different jobs keep on fluctuating over time. One of the major contribution of cloud computing is to avail all the resources at one place in the form a cluster and to perform the resource allocation based on request performed by different users. The services offered by cloud computing are divided into three layers; SaaS (Software as a Services), PaaS (Platform as a Services), and IaaS (Infrastructure as a Services). In SaaS, software application is made available by the cloud provider. With it, the user does not need to buy the software, only rent service business operation computing application infrastructure on the cloud. In PaaS, an application development platform is provided as a service to the developer to create a web based application. With it, the user does not need to buy the software, but the rent service business operation computing application infrastructure on the cloud. This classification enables the users to easily understand and choose the appropriate and suitable cloud services compatible with their business needs. [1] Though cloud computing offers numerous benefits for enormous applications, it comes with certain limitation too. The essence of the technology is to provide best computing facilities to unlimited end-users in reliable, efficient, securable and scalable manner. Cloud Computing is still growing and has many issues and challenges. The main issue is to satisfy the user s requests for better service and using computer resources efficiently in minimum time. Hence scheduling comes into play. There are a large number of virtual servers in a single datacenter running at any instance of time, supporting number of tasks and in the meantime the cloud system continues to take up the batches of task requests. During this, few target servers are needed to be picked out of many available servers, which can fulfill a batch of incoming tasks. It demonstrates task scheduling is a valuable issue which to determine the performance of cloud service provider. It is crucial and extremely important to develop an efficient task scheduling RES Publication 2012 Page 133

2 algorithm to completely utilize the power of cloud computing and to enhance its performance and throughput. Task Schedule is an NP-hard problem. There are generally two levels of Task scheduling in cloud computing, first level is user level that schedule task between service provider and user while the second level is system level that schedule management resource within datacenter. Figure 1: Job scheduling in cloud Numerous algorithms have been proposed and implemented till date such as First Come First Serve, Data-aware algorithm, Min-Min algorithm, Round-Robin algorithm etc., but the optimized scheduling of the individual tasks in cloud is still an issue to solve. Many of the task scheduling algorithms in cloud computing use a single criteria which does not define efficient resource utilization. Hence, it is important to address multiple criteria to enhance the system performance and increase resource utilization. This paper aims to address multiple criteria including load balance, quality of service, economic principles, best running time and throughput by proposing a multi objective task scheduling scheme using Quality of Service parameters of resource nodes. II. REVIEWED WORK There are numerous algorithms proposed by scholars for scheduling that provide better resource scheduling in cloud environment. Scheduling is assigning fundamentally number of resources to the tasks in a manner that there will be maximum utilization of resource, minimum time for total processing and minimum time of waiting. In this section study and review of various load balancing and scheduling algorithms of various authors has been done on the basis of various performance parameters like bandwidth, cost, deadline, execution time, make span, priority, reliability, scalability, time, and throughput. A.I.Awad et.al [1] addresses the dynamic Multi-objective task scheduling in Cloud Computing using an optimized Particle Swarm optimization and presents efficient allocation of tasks to available virtual machine in user level base on different parameters. The paper proposes a mathematical model multiobjective Load Balancing Mutation particle swarm optimization (MLBMPSO) to schedule and allocate tasks to resource. The proposed algorithm considers two objective functions to minimize round trip time and total cost. First objective function is to minimize Expected Round Trip Time (ERTTij) of task i in vm j and the second objective function is to minimize Expected Total Cost (ETCij) of task i in vm j. The weighted sum approach is used to solve multi-objective problem. A multi-objective task scheduling algorithm is proposed by Ekta S. Mathukiya et.al [2] that performs non-dominated sorting for ordering of tasks. Task size, makespan, and deadline are considered as the criteria in the proposed algorithm. The task s priority here is allocated in accordance with the QoS value and QoS are assigned to the VMs on the basis of their Millions Instructions persecong (MIPS) value. The list of VMs possessed by cloud broker is updated after fixed time interval. Based on MIPS the list of VMs is sorted in descending order starting from high QoS VM to low QoS VM and non- dominated sorting is performed on the list to generate non-dominated task s set. These tasks are bounded to VMs sequentially and the process of allocation is repeated for all tasks. Atul Vikas Lakra et.al [3] explained the underutilization of cloud resources due to poor scheduling of tasks in cloud and came up with a multi- objective task scheduling algorithm in this regard to optimize the throughput, cut the execution expenses at the same time keeping the SLAs intact. The proposed scheduling technique prioritizes the tasks according to the QoS such that low QoS valued tasks are assigned higher priority and vice versa. Cloud broker assigns QoS to VMs on receiving list of VMs from cloud service providers. Sandeep Singh Brar et.al [4], came up with a Max-Min algorithm for optimizing workflow scheduling. In this approach, different types of workflows are assigned as an input. The Four wide used scientific workflows used as input are: Montage, Cyber Shake, and SIPHT. Then, different scheduling algorithms like FCFS, Cyber Shake, Sipht and MaxMin are applied and executed. The results showed that Max- Min produced different results for different inputs. A modification of Improved Max-min task scheduling algorithm has been proposed by UpendraBhoi [5]. The proposed enhanced version of Max-min also includes the expected execution time as a selection basis just like the old improved Max-Min method but there is a slight difference in RES Publication 2012 Page 134

3 their working that differentiates them vividly. The already used Improved Max-min algorithm assigns a largest task (having Maximum execution time) to a slowest resource (one producing Minimum completion time) whereas the Enhanced Max-min assigns the task having average execution time to the slowest resource (producing Minimum completion time). One such effort is also done by Navdeep Kauret.al [6] to optimize conventional Max- Min algorithm. It involves the principle of sorting jobs (cloudlets) based on completion time of cloudlets. The traditional Max-Min allocates the longer jobs to better resources and reduces the overall task runtime. So, the overall focus is on time parameter, but factor like storage capacity lacks the consideration. Unlike conventional Maxmin algorithm, the improved algorithm works on multiple parameters. In addition to the completion time parameter, the improved algorithms also considers the storage requirement into focus as this algorithm is being optimized for data intensive specific application. The new algorithm designs weight to each parameters based on its proportion of contributing as a resource for excluding a particular jobs. The completion time statistics are based on a particular virtual machine. P=T α / R β, where T is the payload storage rate potentially achievable for a particular virtual machine, R denotes the heuristic average of payload storage of a particular virtual machine and α, β denotes the time the fairness schedule. The balance between serving best virtual machines can be achieved by changing the α, β values. The completion time can be calculated along with equation: P=T α /R β +CT All this is done, for the job having maximum size as it is based in Max-Min algorithm. It is also called Prioritization coefficients. Hong Sun et.al [7] emphasized on task scheduling algorithms based on comprehensive QoS. The paper considers the new Berger s Model under dynamic cloud computing environment and relates to benefit- fairness algorithm. New Berger game theory model apply the theory of social distribution and game theory on task scheduling in cloud computing environment. In cloud, task scheduling is achieved by assigning independent task to m virtual machine resource to fully utilize the resources in minimum finishing time. If FTi is the finish time of task i, then span is defined as FT max = max(fti, i=1,2, 3..n) The scheduled task is to find the optimal collection that make the spans FT max and ΣFTI minimum in the 2m subset of the possible resource space. FT max = max(fti, i=1,2, 3..n) The scheduled task is to find the optimal collection that make the spans FT max and ΣFTI minimum in the 2m subset of the possible resource space. Huankai Chen et.al [8] took min-min algorithm into consideration. The authors have studied the algorithm and came up with the pros and cons associated with it. Though it is a simple, efficient algorithm and produces a better schedule that minimizes the total completion time of tasks as compared to other algorithms in the literature but the biggest drawback and central issue for cloud providers is load imbalanced. Hence an improved load balanced algorithm LBIMM is proposed in this paper to optimize the load balance of Min- Min to increase the utilization of resources with light load or idle resources thereby freeing the resources with heavy load. Secondly, user-priority aware is introduced PA-LBIMM so as to observe the promised guarantees that VIP customers can enjoy better service than the other ordinary users. KobraEtminani et.al [9], introduced a new scheduling algorithm based on two conventional scheduling algorithms, Min-Min and Max-Min, to use their advantages and overcoming their drawbacks at the same time. It selects between the two algorithms based on standard deviation of the expected completion time of tasks on resources. The proposed scheduling heuristic, the Selective algorithm, has been evaluated within a grid simulator called GridSim and its performance is compared with its two heuristics. The results show that the new algorithm leads to performance gain. RES Publication 2012 Page 135 III. JOB SCHEDULING MODEL IN CLOUD When more than one processes starts to run in a computer simultaneously, competing with each other for the CPU resources then Operating System has to decide in that situation that which process to run next. This process of taking decision that which process will get resources is called Scheduling. Similarly, in cloud computing too, task scheduling is a vital part. It is achieved by scheduling subtasks of workflows by arranging the tasks in queue and providing them with the suitable resources that can execute these tasks. Cloud Computing has many features like virtualization and flexibility. With the help of technology of virtualization, all the resources that are physically available can be made virtualized and transparent for users. [4] The main motive of task scheduling is to attain better cloud performance in terms of better throughput, load balancing among resources, quality of service (QoS), economic feasibility and the optimal operation time. Task scheduling can be viewed from two directions- from the cloud resources users view, users have to identify which cloud computing resource can meet their job QoS requirements for computing and how

4 much amount to be paid for the cloud computing resources. While, from the cloud computing service providers view, its concern is to gain the maximum profits by offering cloud computing resources, apart from meeting the CCU s job QoS requirements. Load balancing and job scheduling has close contacts with each other in the cloud environment. Because of the pertinency of job scheduling algorithm, load balancing become another important measure in the cloud. Load Balancing is a technique which divides the workload across multiple computing resources such as computers, hard drives and network. In this fair allocation of resources of client request tried to achieve in the best way to ensure proper utilization of resource consumption. In a Cloud computing environment different load balancing scheduling exists among which first, is the Batch mode heuristic scheduling algorithm where jobs are queued in a set and collected as batches as they arrive in the system after which they get started after a fixed time period. Example: First Come First Serve (FCFS), Min- Max algorithm, Min-Min algorithm and Round Robin (RR) algorithm. Second, one is On-line mode heuristic scheduling here jobs are scheduled individually as they arrive in the system. Several algorithms & protocols have been proposed till date regarding the scheduling mechanism of the cloud computing. Most of the traditional scheduling strategies were singleobjective which could focus on one particular issue. The main examples of traditional scheduling algorithms are FCFS, Round-Robin, Min-Min algorithm, Max-Min algorithm and meta- heuristic algorithms. In last few years, multi- objective strategies have been explored and developed. Multiple criteria include improving the various Quality of Services and at the same time maintaining the efficiency and fairness among the jobs. IV. PROBLEM FORMULATION The primary goal of Cloud Computing is to provide efficient access to remote and geographically distributed resources with the help of Virtualization in Infrastructure as a Service (IaaS). Various virtual machines (VM) are needed as per the requirement and cloud provider provides these services as per the Service Level Agreement (SLA) to ensure QoS. For managing enormous amount of VM requests, the cloud providers require an efficient resource scheduling algorithm. The traditional or any rule-based scheduling algorithms are widely used on cloud computing systems because they are simple and easy to implement, even provide optimal solution for scheduling problem, but they are all suffered from the fact of not being able to find the optimal solution in a reasonable time, for too complex or large problems. In the existing approach, cloudlets were being scheduled according to First Come First Serve algorithm after sorting them according to Non-dominated sort technique, thereby mapping the tasks to the virtual machines, according to their arrival times. But the concept of balancing the load of virtual machines was not there, which was a big drawback to the existing system. The paper aims to study and analyze the processing time of low level scheduling algorithms and to develop a multiobjective task scheduling using quality of service parameters of resource nodes. It also attempts to improve the performance of CPU, memory and network operations by reducing the load of a virtual machine (VM) by using Load Balancing Method. Finally, it optimizes the resource utilization by using Resource Aware Scheduling Algorithm. V. PROPOSED WORK In the proposed architecture, after performing the Nondominated sort, the concept of Task-Based Load Balancing is being taken into consideration. Non-dominated sorting is used to solve the multi-objective problems which target multiple objective functions. In the proposed work, the main goal is to minimize the processing time of a task. As balancing the load of machines helps in distributing the work load of a task onto multiple computers, and it also provides reliability in particular tasks. This algorithm achieves the system load balancing by transferring only extra tasks from an overloaded VM rather than migrating the entire overloaded VM. Moreover, Resource Aware scheduling algorithms (a combined approach of Max- Min and Min-Max algorithms), has also been used to achieve efficient resource utilization, further optimizing the resources in terms of accuracy and efficiency. Small tasks are executed by using Min-Max strategy before the large ones. Max- Min strategy comes into play to avoid delays in the execution of large tasks, to support concurrency in the execution of large and small tasks. So, these approaches are considered to give us a way better results than the previous scenarios. A. Min-Min The Min-Min algorithm is simple many cloud scheduling algorithms are based on it. It begins with a set S of all unmapped tasks. Then the resource R which has the minimum completion time for all tasks is found. Next, the task T with the minimum size is selected and assigned to the corresponding resource R. Last, the task T is removed from set S and the same procedure is repeated by Min-Min until all tasks are assigned. This algorithm fails to utilize the resources efficiently which lead to a load imbalance. And without use-priority aware during scheduling, the VIP users are not guaranteed with better RES Publication 2012 Page 136

5 services which lead to VIP users dissatisfaction. For a set of i tasks (T1, T2, T3 Ti) to be scheduled onto j available resources (R1, R2, R3 Rj). Min- Min algorithm is explained as: 1.for all tasks Ti in meta-taskm v 2. forall resources Rj 3. Cij=Eij+rj 4.do until all tasks inmv are mapped 5. for each task in Mv find the earliest completion time and the resource that obtaines it 6. find the task Tk with the minimum earliest completion time 7. assign task Tk to the resource Rl that gives the earliest completion time 8. delete task Tk from Mv 9. updaterl 10. update Cij for all i 11.end do B. Max-Min The Max-min algorithm is commonly used in distributed environment and it begins with a set of unscheduled tasks. Then expected execution matrix and expected completion time of each task on the available resources is calculated. Further the task with overall maximum expected completion time is chosen and assigned to the resource with minimum overall execution time. Finally recently scheduled task is removed from the metatasks set, updating all calculated times, then repeating until meta-tasks set become empty. Max-min algorithm losses some of its major advantages as load balance between available resources in small distributed system configuration and small total completion time for all submitted tasks in large scale distributed environment. C. RASA RASA is a combined approach of Max-Min and Min-Max algorithms that performs efficient resource utilization, further optimizing the resources in terms of accuracy and efficiency. This algorithm uses both the algorithms Max- min and minmax to use the pros of both and eliminate the drawbacks. To realize this optimization, it first estimates the completion time of the tasks on each of the available cloud resources and then applies the Max-Min and Min-Min algorithms, alternatively. Small tasks are executed by using Min-Min strategy before the large ones. Max- Min strategy comes into play to avoid delays in the execution of large tasks, to support concurrency in the execution of large and small tasks. The key objective of the proposed algorithm is to obtain better results than existing algorithm of Shortest Job First (SJF)to reduce the execution time of jobs. The evaluation parameters considered for performance evaluation are: a) Average Waiting Time b) Total Processing time c) Total Processing cost RASA with Max- Min and Min- Min algorithm is as follows: 1. for all tasks Ti in meta-taskmv 2. forall resources Rj 3. Cij=Ej+rj 4. do until all tasks in Mv are mapped 5. if the number of resources is even then 6. for each task in Mv find the earliest completion time and the resource that obtains it 7. find the task Tk with the maximum earliest completion time 8. assign task Tk to the resource Rl that gives the earliest completion time 9. delete task Tk from Mv 10. Updaterl 11. Update Cij for all i 12. else 13. for each task in Mv find the earliest completion time and the resource that obtains it 14. find the task Tk with the minimum earliest completion time 15. assign task Tk to the resource Rl that gives the earliest completion time 16. delete task Tk from Mv 17. updaterl 18. updatecil for all i 19. end if 20. end do RES Publication 2012 Page 137 VI. METHODOLOGY 1. Develop multi objective task scheduling. In this, tasks are sorted using non-dominating sort and then the sorted tasks are mapped to virtual machines in order to optimize the processing and average waiting time. Non-dominated sorting (task list) i=0 Create empty non-dominated list Initially put task (i) into non-dominated list

6 For all i=1 to size of task list For all j=0 to size of non-dominated list If task (j) dominates task (i) i.e. checking the task length and task Output file size Then Put task (j) into non-dominated set Else if task (i) dominates task (j) Then Put task (i) into non-dominated set Else Put task (i) and task (j) into non-dominated set End if End for End for 2. Load balance of virtual machines is achieved by first mapping tasks to VM s and then all the VM to host resources, using the Task-Based System Load Balancing Method. This algorithm ensures the system load balancing through only transferring extra tasks from an overloaded VM instead of migrating the entire overloaded VM. The loads are formulated as: By taking these parameters for calculating the load, we will get the better balanced load virtual machine for the task to be performed on cloud. n i=1 cloudlet_length ij Fitness1 ij = Vm j _mips Where Vmj_mips is defined by millions of instructions per second for each processor of Vm j, n is the total number of scout foragers, fit ij defines the fitness function of machines (i) for Vm j or say capacity of Vm j withi th virtual machine number, cloudlet_length is defined as the task length that has been submitted to Vm j. The virtual machine (Vm j ) capacity is being calculated using the following parameters Capacity_Vm j = Vm j _cpu Vm j _size + Vm j _bandwidth Where Vm j _cpu the number of processors in a virtual machine Vmj is, Vm j _size is the virtual machine memory size, Vm j_ bandwidth ability of Virtual machine Vm j. then applies the Max-Min and Min-Max algorithms, alternatively. Small tasks are executed by using Min-Max strategy before the large ones. To avoid delays in the execution of large tasks to support concurrency in the execution of large and small tasks Max-Min strategy is used. 4. Scheduling strategies considering parameters are : a) Total Processing time b) Total Processing Cost c) Average Waiting Time Formulation of three of the parameters is given below: Parameters Total Processing Time Total processing Cost Average Waiting Time Formulation CloudletLength / vmmips*vmnumberofpes characteristics.getcostpermem * vm.getram cloudlet.waitingtime Fitness2 ij = n i=1 cloudlet_length ij Vm j _Size VMload = Mean of (Fitness1+Capacity_vm+Fitness2) LowloadedVM = maxcap-(load/capacity) OverloadedVM = (load/capacity) maxcap 3. Resource utilization is achieved by using RASA a combined approach of Max-Min and Min-Max algorithms to further optimize the resources in terms of accuracy and efficiency. To achieve this, it first estimates the completion time of the tasks on each of the available cloud resources and RES Publication 2012 Page 138

7 VII. FLOWCHART Initialize CloudSim by specifying number of users, flag and Calendar reference Create Datacenters, virtual machines and cloudlets Sort cloudlets using nondominating sort by cloudlet length, file size and File Output size of the cloudlet Perform Task Based Load balancing VIII. SIMULATION ANALYSIS This section presents the simulation results of the proposed methodology implemented with the help of Cloudsim and Net beanside8.0. The proposed scheme is evaluated using different scenarios where varying number of jobs are assigned to virtual machines. In different scenarios the load balancing capacity is analyzed. Following table presents the summarized view of the different scenarios taken into consideration along with the parameters of performance analysis. The performance of the proposed scheme is evaluated against the previous approach and the comparative analysis is described. Many VMs and tasks with different task size have been created. Task size ranges from 100 to 500. Table 1: Showing the end result analysis of the proposed and the previous scheduling model in different scenarios If Load<Threshold LowLoaded VM If Low Loaded VM found Perform Min- Max scheduling i.e sending Perform RASA OverLoaded VM If Over Loaded VM found Perform Max- Min scheduling i.e sending Virtual Cloud Environ ment V M s Job s Total Processing Time (ms) Previous E+0 7 Proposed E+0 7 Performance Analysis Total Processing cost ($) Previou s Propos ed Average Waiting Time(ms) Prior New Evaluate performance parameters like waiting time, processing time and processing cost Table 1 shows the performance analysis of the proposed technique with the existing technique on the basis of different tasks mapped to different number of virtual machines. The table contains the comparison of three parameters including total processing time, average waiting time and total processing cost of the proposed and the existing approach. The analysis is done by taking less number of machines and more number of cloudlets in order to show the load balancing concept in best manner i.e. how the large number of task has been scheduled to the virtual machines. From the analysis of the table it is cleared RES Publication 2012 Page 139

8 shows that the proposed approach performs better in all the mentioned parameters like processing time, processing cost and average waiting time. The performance analysis is further illustrated graphically: Figure 3: Graphical Representation for Total Processing Time of Previous and Proposed model in different scenarios IX. CONCLUSION In this paper, Multi- objective task scheduling is emphasized and to achieve the same a multi- objective scheduling algorithm is proposed considering the user s Quality of Service requirements. The proposed algorithm considers three parameters i.e. Total processing cost, total processing time and average waiting time. Load balancing technique is used to guarantee balancing of load and RASA is used for proper resource utilization. RASA is hybrid approach of Max-Min and Min-Min algorithms. The parameters considered in this research work are tested with various input values and the results depicts that the proposed method achieves better results than the existing methods. The work can be further extended in future aiming to achieve more efficient performance results. The proposed work is using RASA for resource allocation. In future work, resource allocation can be performed with the improved versions and evolutionary algorithms. In addition to this, the proposed system can be further implemented in real time scenario. ACKNOWLEDGMENT The paper has been composed with the kind assistance, guidance and support of my department who have helped me in this work. I would like to thank all the people whose encouragement and support has made the fulfilment of this work conceivable. Figure 4: Graphical Representation for Total Processing Cost of Previous and Proposed model in different scenarios Figure 5: Graphical Representation for Average Waiting Time of Previous and Proposed model in different scenarios REFERENCES [1] Tarek Z, Zakria M,.Omara F A, PSO Optimization algorithm for Task Scheduling on The Cloud Computing Environment, ISSN , International Journal of Computers And Technology Vol. 13, No. 9 [2] Mathukiya E.S, Gohel P.V, Efficient Qos Based Tasks Scheduling usingmulti-objective Optimization for CloudComputing, International Journal of Innovative Research in Computerand Communication EngineeringVol. 3, Issue 8, August 2015 [3] Lakra.A.V,Yadav.D.K, Multi-Objective Tasks Scheduling Algorithm for Cloud Computing Throughput Optimization, International Conference on Intelligent Computing, Communication & Convergence, 2015 [4] Brar S. S., Rao S., Optimizing Workflow Scheduling using Max-Min Algorithm in Cloud Environment, International Journal of Computer Applications ( )Volume 124 No.4, August 2015 [5] Bhoi U, Ramanuj P.N, Enhanced Max-min Task Scheduling Algorithmin Cloud Computing, International Journal of Application or Innovation in Engineering and Management, Volume 2, Issue 4, April 2013 RES Publication 2012 Page 140

9 [6] Kaur N., Kaur K, Improved Max-Min Scheduling Algorithm, IOSR Journal of Computer Engineering (IOSR-JCE)e-ISSN: ,p-ISSN: , Volume 17, Issue 3, Ver. 1 (May Jun. 2015), PP [7] Sun H, Chen S.P, Jin C, Guo K, Research and Simulation of Task Scheduling Algorithmin Cloud Computing, TELKOMNIKA, Vol.11, No.11, November 2013, pp. 6664~6672e-ISSN: X [8] Chen H, Wang F, Dr Helian N, Akanmu G, User-Priority Guided Min-Min Scheduling Algorithmfor Load Balancing in Cloud Computing [9] Etminani K, Naghibzadeh M, A Min-Min Max-Min selective algorithm for grid task scheduling, DOI: /CANET Source: IEEE Xplore [10] Liu J, Luo X. G, Zhang X.M, Zhang F and Li B.N, Job Scheduling Model for Cloud Computing Based on Multi- Objective Genetic Algorithm, IJCSI International Journal of Computer Science Issues, Vol. 10, Issue 1, No 3, January 2013 [11] Pandey S, Wu L, Guru S M, Buyya1R, A Particle Swarm Optimization-based Heuristic for Scheduling WorkflowApplications in Cloud Computing Environments [12] Tripathy L, Patra R.R, Scheduling In Cloud Computing, International Journal on Cloud Computing: Services and Architecture (IJCCSA) Vol. 4, No. 5, October 2014 [13] Agarwal A, Jain S, Efficient Optimal Algorithm of Task Scheduling in CloudComputing Environment, International Journal of Computer Trends and Technology (IJCTT) volume 9 number 7 Mar 2014 [14] Ghanbaria S, Othman M, A Priority based Job Scheduling Algorithm in Cloud Computing, International Conference on Advances Science and Contemporary Engineering 2012(ICASCE 2012) [15] Vijayalakshmi M, Muthusamy K, An Efficient Study of Job Scheduling Algorithmswith ACO in Cloud Computing Environment, Volume 3, Special Issue 3, March International Conference on Innovations in Engineering and Technology [16] B. Kanani and B. Maniyar, Review on Max-min Task Scheduling Algorithm for Cloud Computing, Journal of Emerging Technologies and Innovative Research,vol. 2, pp ,2015 [17] A.Bhatia and R.Sharma, An Analysis Report of Workflow Scheduling Algorithm for Cloud Environment, International Journal of Computer Applications, vol.119, pp ,2015 [18] Kowsik, K. RajaKumari, A Comparative Study on Various Scheduling Algorithms in Cloud Environment, International Journal of Innovative Research in Computer and Communication Engineering, vol.2, 2014 Master of Technology in Computer Science at SVIET. Her major areas of interest are Cloud Computing, Data Mining and Software Engineering. She has 4 publications in reputed journals. Vikas Zandu is an assistant Professor in SVIET. He has completed his B. Tech and MTech in Computer Science in 2011 and 2013 respectively. His main areas of interest is Networking. He has a numbers of publications in good journals and conferences. AUTHOR S BIOGRAPHIES Kavita is a student of Swami Vivekananda Institute of Engineering and Technology, Punjab. She completed her Bachelor of Technology degree in Computer Science and Engineering from Lovely Professional University in She is now pursuing RES Publication 2012 Page 141