A Survey on Generating Load Balancing Algorithm for Cloud Computing

Abstract

The elaboration of IT led Cloud calculating technology crop as a new prototype in furnishing the services to its druggies on rented base at any time or place. Considering the inflexibility of cloud services, innumerous associations switched their businesses to the cloud technology by setting up more data centers. Nonetheless, it has come obligatory to give profitable prosecution of tasks and applicable resource application. A many approaches were outlined in literature to enhance performance, job scheduling, storehouse coffers, QoS and cargo distribution. Cargo balancing conception permits data centers to fore stall over-loading or under- lading in virtual machines that as similar is an issue in cloud computing sphere. Accordingly, it bear the experimenters to layout and apply a proper cargo balancer for cloud terrain. The separate study represents a view of problems and pitfalls faced by the current cargo balancing ways and make the experimenters find more effective algorithms. cargo unbalancing problem is a multi-variant, multi-constraint problem that degrades performance and effectiveness of computing coffers. cargo balancing ways feed the result for cargo unbalancing situation for two undesirable angles- overfilling and under- lading. In disdain of the significance of cargo balancing ways to the stylish of our knowledge, there\'s no comprehensive, expansive, methodical and hierarchical bracket about the being cargo balancing ways. Further, the factors that beget cargo unbalancing problem are neither studied nor considered in the literature.

Introduction

I. INTRODUCTION

Cloud cargo balancing is defined as the system of splitting workloads and calculating parcels in a cloud computing. It enables enterprise to manage workload demands or operation demands by distributing coffers among multitudinous computers, networks or waiters. cloud cargo balancing includes holding the rotation of workload business and demands that live over the Internet.  As the business on the internet growing fleetly, which is about 100 annually of the present business?

The cloud computing can be described as an on- demand service pool which connects colorful waiters to each other for furnishing services to aiming guests. The cloud providers may contain direct access to these services. thus, the coffers can be used according to the demand. The stoner can prize and modifies the data stored in the shadows. The different services to the stoner are handed on demand using a point called “ cloud service provider ”. This particularity makes certain that the quantum of services being employed for any number of times can be employed for calculating the expenditure of the stoner to pierce that service. The cloud calculating system provides extremely complicated operations in different surroundings. In addition, some professed concentrated services are handed in each terrain. In cloud computing, common group of coffers is handed to the druggies. Using cloud computing, the druggies can use these coffers according to their need far and wide. The main ideal of this technology is to maintain the minimal cost to pierce the services. It's anatomized that the software and tackle means attained using internet remain present in the virtual system and supports to give the services. The stoner accesses a common group of coffers using cloud computing on the base of demand. The virtualization allows stoner to subscribe and use the services for a certain time period by getting access of the common group of coffers using cloud computing.

Load balancing results can be distributed into two types –

1. Software- grounded cargo balancers Software- grounded cargo balancers run on standard tackle( desktop, PCs) and standard operating systems.
2. Tackle- grounded cargo balancer tackle- grounded cargo balancers are devoted boxes which include Application Specific Integrated Circuits( ASICs) acclimated for a particular use. ASICs allows high speed promoting of network business and are constantly used for transport- position cargo balancing because tackle- grounded cargo balancing is briskly in comparison to software result.

II. LOAD BALANCING MODEL BACKGROUND

In cloud computing, cloud waitpersons should always be balanced, to use the resources with their full capacity. sometimes it happens that some waitpersons are heavily loaded while the other waitpersons are under loaded or in idle state. To overcome this problem weight balancing algorithms are used. These algorithms help in allocating every single task by covering weight on each garçon. According to the balancing algorithm is defined as “ The weight balancing in murk may be among physical hosts or VMs. This balancing medium distributes the dynamic workload inversely among all the bumps( hosts or VMs). The weight balancing in the cloud is also appertained to as weight balancing as a service( BaaS) ”.

 There are different types of weight balancing algorithms which are used for cloud computing; they are categorized in two orders videlicet stationary weight balancing and dynamic weight balancing. stationary weight balancing algorithms allocate tasks to the waitpersons before the florilegium where all the conditions of the resources are known to the algorithm. The allocation of tasks are predicated on those conditions. stationary weight Balanced Algorithm is suitable for small distributed surroundings with high internet speed and ignorable communication detainments It works properly when the systems or bumps have the ignorable differences in the weight, therefore the algorithms which come under the static are generally not suitable for cloud computing. Because in the cloud we have n number of stoners due to which weight largely varies. Dynamic algorithms work in the real time situation, where it takes continuous information about the weight on the garçon. With respect to that it takes the decision of distributing the tasks amongst the waitpersons. Accordingly we can allocate, reallocate or remove any task from the garçon predicated on the priority. Dynamic weight Balanced Algorithm focuses on reducing communication detainments and execution time for large distributed surroundings. These ways or the algorithms are largely successful for weight balancing the cloud terrain on their bumps among different types of resources. In the formerly numerous times there are multitudinous static and dynamic weight balancing algorithms that have been proposed for the cloud calculating terrain. A detailed comparison is done in. In this section some of the being algorithms proposed by the researchers are mooted. The static weight balancing algorithm factory in small distributed surroundings so they are less complex compared to dynamic which have a largely distributed terrain. Advanced information is demanded in static algorithms analogous as length and number of tasks. The scheduling opinions are taken at runtime by dynamic algorithms and collect time by static. stationary algorithms are not good at balancing weight properly at run time but monitor bumps continuously where dynamic balance loads efficiently and do the monitoring continuously by event base or time interval. stationary algorithms take farther time to break but do not give the optimal result for the complex computational problem, dynamic takes lower time and gives useful results. Traditional types of algorithms come under the stationary bones and the meta- heuristic algorithms come under the dynamic algorithms.

III. RELATED WORK

However, load balancing has been an eagle’s eye among researchers because of its essence in cloud computing between the stakeholders’ i.e. Cloud Service Provider and Cloud Service Consumer. Based on analysis of existing review literature one of the reasons presented is absence of proper classification among different approaches. A thorough review about the existing work in literature has been presented in this section.

Sovban Nisar, Deepika Arora [1] For handling the problems related to node failure in cloud networks, an algorithm named BFO is used in this research. Several nodes are included in a proposed algorithm. Depending upon the failure rate and minimal execution time, a participant node is chosen among all these nodes. In this scenario, the threshold value is fixed using the master node. There are two parameters included in this threshold value. The master node chooses nodes having equivalent or less failure rate with least execution time as the participant nodes. In comparison to threshold value, the value of node N1 is less.

Mohieddin Harb [2] As a result of the drawback of using throttled load balancing algorithm we proposed the balanced throttled load balancing which work as follow: The index table of all the virtual machines is maintained by Balanced Throttled Load Balancer. This also maintains the state of each virtual machine i.e. whether the virtual machine is busy or available. Initially, at the start of the algorithm, all the virtual machines have been present. Then, Data Center Controller gets the fresh task.

Sambit Kumar Mishra [3]  The task allocation algorithms in the cloud are classified based upon the current state of VM. In allocation policy where the current load information of VMs are available before the allocation is said to be a dynamic strategy. Whereas the static strategy acts on VMs without any load information. Load balancing attends in fair allocation of resources to achieve a high user satisfaction and improve the stability of the system. We have proposed a taxonomy for the load balancing algorithms in the cloud environment as shown in Fig. 3. Resource management plays a major role in the load balancing of cloud resources.

Dr. Sharvani GS [4] With more and more advances being made in cloud computing and its increasing efficiency, companies have started using cloud as their underlying architecture for most of the important operations. The demand for resources is always increasing in these companies and with the help of cloud architecture, all the demand requirements are met easily. Cloud allows them to increase/decrease the load on servers according to their requirements as cloud provides the policy of pay-as-you-go which makes it a good option for the organizations.

Iehab AL Rassan and Noof Alarifi [5] proposed a fine-grained data access control withattribute hidingpolicy  for  cloud-based  IoT. A fine-grained access control policy was also put forward to support an excessive  access  policy  with  full  attributes  hidden  for cloud-based IoT. Herein, attribute-based information is fully hidden using  arandomizable  technique. A  fuzzy  attribute positioning  mechanism  is  used  to  locate  the  attributes  of authorized users efficiently. A garbled bloom filter is used for this process.  However, the study’s use of the garbled bloom filter causes a high number of false positives, which indicate that an attribute is a member of an access policy group when it is really not.

Muhammad Asim Shahid [6] LB provides a systematic mechanism for the equal distribution of the responsibility to the resources available. The goal is to provide reliable service, including adequate use of the resource, in the event of a disaster of the portion of any service by supplying & de-provisioning the device instance. In addition, LB is aimed at reducing response time for tasks & increasing resource ef_ciency, which increases device ef_ciency at a lower cost.

Asha Sohal, Ramesh Kait [7] Cloud computing is a wide and fastest growing area in terms of computing research and industry these days. It mainly provides services based on IaaS, SaaS, and PaaS. These are the key parameters which decide the role of cloud services to the end users. These services can be offered to the end users through virtualization over the internet. Cloud has many advantages like large scaled computing, flexible infrastructures, pay per use, on demand services and many more. There are some major issues in processing of jobs over cloud computing like security, equal distribution of load, fault tolerance etc. and the biggest challenge over cloud is latency time which means the total time between the data sent by IoT over cloud, processing time and finally reply to the IoT or vice versa.

Sally F. Issawi [8] The request rates received by the data center are not constant all the time. Sometimes large number of requests aggregated in a small period of time creating a burst. This affect the performance of the load balancing algorithm as it increase the processing time and the repose time of the data center. The performance of several load balancing algorithms differs according to the users’ requests rate. For example some algorithms work efficiently under low workload while their performance is degraded under high workload and vice versa. To overcome burst problem and benefit from different load balancing algorithms advantages we propose a new load balancing algorithm called Adaptive algorithm.

Sovban Nisar[9] The evolution of Virtualization, Utility computing, Software-as-a-Service (SaaS), Infrastructure-as-a-Service (IaaS) and Platform-as-a-Service (PaaS) all are combined to make a cloud computing and three development models of cloud are public, private and hybrid. Public cloud services are available for general public over the internet . Private cloud is used for personal use or provides services to single organization. A hybrid cloud is combination of two or more than two public and private cloud which are bounded by service level agreement (SLA). Clients/Users can forward the requests at any time from any geographical location/region for the required services, SLA selects the best resource within user defined deadline and budget. Elastic resource provisioning with quality of service (QoS) parameter (deadline, high availability, priority etc.).

Yelchuri Venkata Sai Harsha[10] Cloud computing is the distribution of diverse offerings along with storage, servers, networking, software programs, intelligence, and analytics, through the internet so as to offer faster innovation, more flexible sources, and economies of scale. Take an illustration of a site open to everyone. A high number of clients can visit a site or online application whenever. A web application's capacity to deal with these client demands without a moment's delay gets intense. It might even cause system failures. The terrible sense of a website being down or not accessible also delivers lost prospective clients for a website owner whose entire career is based on his portal. Load balancing is crucial in this situation.

IV. SYSTEM MODEL

In cloud computing there are many load balancing algorithms. The most used load balancer algorithms are: Round Robin Algorithm (RR), Active Monitoring Load Balancing algorithm (AMLB) and Throttled Algorithm (TLB) [4]. “Cloud Analyst Simulator” is used to evaluate the used algorithms.

A. Load balancers Algorithms

1. Round Robin Algorithm (RR)

It's a simple way to distribute requests from guests across waiters. A customer request is encouraged to each garçon in turn. It's the most habituated algorithm, and it's easy to apply and conceptualize.

 In this algorithm, requests from guests are routed cyclically to the available waiters. It works good when waiters have roughly the same computing and storehouse capabilities.

2.  Active Monitoring Load Balancing Algorithm (AMLB)

It maintains a virtual machine indicator table and the number of allocations allocated to each virtual machine. Data Centre Controller receives a new task from a customer. When a request from Data Centre Controller arrives at AMLB, The AMLB scans the indicator table from top. When it finds the least loaded VM, AMLB returns VM id to data centrecontroller.However, AMLB selects the first linked, If there are further than one loaded VM. The data centre regulator sends the request to the named VM.

3. Throttled Load Balancing Algorithm (TLB)

In this algorithm the cargo balancer responsible for preparing indicator table. It displays the information of the virtual machine state either Available or Busy. When the task is arrived, the cargo balancer assigns the task to suitable virtual machine. Which is available to execute the stoner task( Fig 4 Developed from( 9)). But every time strangled checks the table from first indicator to determine the available virtual machine.

V. EXPERIMENTAL STUDY

Regarding the experimental study execution, the preferential metric chosen for the model is energy consumption values. Thus, the metric for calculation covered in the system is for “Le” for a make-span. The data in terms of units of energy consumed for the time frame T is considered for the experimental study as Instance-1 (depending on the cloud environment and load factors, the instance-1 could be presumed for different timelines. For experimental purpose, it is seen as an instance, and undefined timeframe). The following stands the 50 instance values generated over the random value of energy consumption for a specific cloud server. The first 20 closing values are used for generating the random values for two virtual machines as a comparison. Considering the scope of experimental analysis, the model refers to the conditions wherein the Le values are chosen for two different Virtual machines for a simple figurative representation of the impact. In cloud computing, load balancing is required to distribute the workload evenly across all the nodes. Using a proper load balancing algorithm leads to minimizing the resource consumption and over provisioning of resources, which improves the performance of the system in different aspects and helps to achieve higher user satisfaction. In this paper, we propose a method that combines the advantages of Round robin with a threshold technique. Thus, the proposed scheduler enhances scheduling policies to overcome the resource utilization problem and to reduce the make span of the system. An overview of the environment and its components is presented in the next subsection.

System overview and main components A typical environment for modeling cloud systems usually involves datacenters, set of host machines and a set of virtual machines. Datacenter: is centralized place that involves numerous servers where computing and networking is taking place in the cloud.

? Host: Host machines are connected to the datacenter. Hosts involve set of virtual machines.

? VM: is virtual machine that is applied on a physical machine. Every virtual machine may run different OS than the physical machine and have their own resources.

Conclusion

After using these four load balancing algorithms we conclude that creating new load balancing algorithm is important task in cloud computing where load balancing is complex task in cloud computing. Comparing the above results of simulation using the four algorithms we can conclude that the overall response time of Balanced Throttled load balancing algorithm is better than the overall response time of throttled load balancing and the other algorithms. The critical advantage of the proposed system is its simplicity in the estimations, which do not require high computation times, and the decision-making process can be much simplified for allocation. The simulated experimental study of the model for a two-VM environment refers to the model’s potential to be implemented in large-scale cloud computing solutions. In addition, evolutionary technique, which shall use the visual indicators as the fitness function.

References

[1] Shereen Yousef Mohamed, Mohamed Hamed N. Taha, “ A Proposed Load Balancing Algorithm Over Cloud Computing (Balanced Throttled)” International Journal of Recent Technology and Engineering (IJRTE) ISSN: 2277-3878(Online), Volume-10 Issue-2, July 2021. [2] Sovban Nisar, Deepika Arora. \" Load Balancing Model using Elastic Technique in Cloud Computing.\" International Journal of Engineering and Advanced Technology (IJEAT) ISSN: 2249-8958 (Online), Volume-10 Issue-4, April 2021 [3] Sambit Kumar Mishra,“ Load balancing in cloud computing: A big picture”, Journal of King Saud University – Computer and Information Sciences 32 (2020) 149–158 [4] Shahbaz Afzal, “Load balancing in cloud computing – A hierarchical taxonomical classification” Journal of Cloud Computing: Advances, Systems and Applications https://doi.org/10.1186/s13677-019-0146-7 [5] JialuHao, Cheng Huang, Jianbing Ni, Hong Rong, Ming Xian, Xuemin (Sherman) Shen, “Fine-Grained Data Access Control with Attribute-Hiding Policy for Cloud-Based IoT,” Computer Networks, 2019. [6] N. Alassaf, A. Gutub, S. A. Parah, and M. Al Ghamdi,“Enhancing speed of SIMON: a light-weight-cryptographic algorithm for IoT applications”, Multimedia Tools and Applications, vol. 78, pp.32633-32657, 2019. [7] M. Karthigaiveni, and B. Indrani, “An efficient two-factor authentication scheme with key agreement for IoT based E-health care application using smart card,” Journal of Ambient Intelligence and Humanized Computing, pp.1-12, 2019. [8] A. Dorri, S. S. Kanhere, R. Jurdak, and P. Gauravaram, “LSB: A Lightweight Scalable Block chain for IoT security and anonymity,” Journal of Parallel and Distributed Computing, vol. 134, pp.180-197, 2019. [9] Muhammad Kazim, Lu Liu, Shao Ying Zhu, “A Framework for Orchestrating Secure and Dynamic Access of IoT Services in Multi-Cloud Environments,” IEEE Access, Volume 6, pp. 58619–58633, 2018. [10] Sana Belguith, NesrineKaaniche, Maryline Laurent,AbderrazakJemai, RabahAttia, “PHOABE: Securely Outsourcing Multi-Authority Attribute Based Encryption with Policy Hidden for Cloud Assisted IoT,” Computer Networks, Volume 133, pp. 141–156, 2018 [11] Qinlong Huang, Licheng Wang, Yixian Yang,“DECENT: Secure And Fine-Grained Data Access Control With Policy Updating for Constrained IoT Devices,” World Wide Web, Volume 21, Issue 1, pp. 151–167, 2018. [12] Ming Tao, JinglongZuo, Zhusong Liu, Aniello Castiglione, Francesco Palmieri, “Multi-layer Cloud Architectural Model and Ontology-Based Security Service Framework for IoT-Based Smart Homes,” Future Generation Computer Systems, Volume 78, pp. 1040–1051, 2018. [13] P. Gope, and B. Sikdar, “Lightweight and privacy-preserving two-factor authentication scheme for IoT devices”, IEEE Internet of Things Journal, vol. 6, pp.580-589, , 2018. [14] Cheng-Yu Yang, Cheng-Ta Huang, Ya-Ping Wang, Yen-Wen Chen, “File Changes With Security Proof Stored in Cloud Service Systems,” Personal and Ubiquitous Computing , Volume 22, Issue 1, pp. 45–53, 2018. [15] M. N. Aman, M. H. Basheer, and B. Sikdar, “Two-factor authentication for IoT with location information,” IEEE Internet of Things Journal, vol. 6, pp.3335-3351, 2018. [16] W. Li, L. Liao, D. Gu, C. Li, C. Ge, Z. Guo, Y. Liu, and Z. Liu, “Cipher text-only fault analysis on the led lightweight cryptosystem in the internet of things,” IEEE Transactions on Dependable and Secure Computing, vol. 16, pp.454-461, 2018. [17] Ebrahim A Alkeem, Dina Shehada, Chan YeobYeun, M. Jamal Zemerly, “New Secure Healthcare System Using Cloud of Things,” Cluster Computing, Volume 20, Issue 3, pp. 2211–2229 , 2017. [18] Gebrie, M.T.; Abie, H. Risk-based adaptive authentication for Internet of things in smart home e-Health. In Proceedings of the 11th European Conference on Software Architecture: Companion Proceedings, Canterbury, UK, 11–15 September 2017; pp. 102–108. [19] P. Xu, X. Tang, W. Wang, H. Jin and L. T. Yang, \"Fast and Parallel Keyword Search Over Public-Key Ciphertexts for Cloud-Assisted IoT,\" in IEEE Access, vol. 5, pp. 24775-24784, 2017. [20] F. Al-Turjman, Y. K. Ever, E. Ever, H. X. Nguyen and D. B. David, \"Seamless Key Agreement Framework for Mobile-Sink in IoT Based Cloud-Centric Secured Public Safety Sensor Networks,\" in IEEE Access, vol. 5, pp. 24617-24631, 2017.

Copyright

Copyright © 2022 Gagandeep Kaur. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.