The current cloud computing environment serves many fields but faces limitations such as security, authentication, fault tolerance, load balancing, and availability [1]. There are various advantages of cloud computing, such as virtualization, resource sharing, ubiquity, and utility computing, but there are also critical issues like security, privacy, load management and fault tolerance [2]. Load balancing is one of the main challenges in cloud computing, as it is required to distribute the dynamic workload across multiple nodes to ensure that no single node is overwhelmed [2]. This paper studies eleven load-balancing algorithms, and various parameters are used to check the results.

Load balancing is essential in cloud computing to ensure an efficient and fair allocation of computing resources. This section reviews existing load-balancing techniques and compares them to a new efficient Virtual Machine (VM) load balancing algorithm [9]. The proposed algorithm finds the expected response time of each resource. It sends the ID of a VM having a minimum response time to the data centre controller for allocation to the new request [10]. The experimental result compares the proposed VM load balancing algorithm with the throttled and active VM load balancers [11]. Researchers proposed the Weighted Active Monitoring Load Balancing (WALB) Algorithm, which creates VMs of different processing power and allocates weighted counts according to computing power [12].

Table 1 compares the reviewed algorithms in terms of the challenges discussed. When a request to allocate a VM arrives, the algorithm identifies the least loaded and most powerful VM according to the weight assigned and returns its VM, ID to the data center controller [13]. The experimental result showed that the proposed algorithm achieves better performance factors such as response and processing time but does not consider process duration for each request [14]. Investigators proposed a new VM load balancing algorithm, modified WALB algorithm, which creates VMs of different processing power and allocates weighted count according to the computing power of the VM enhanced load balancing to avoid deadlock proposed a technique to avoid deadlock among VMs while processing a request by migrating the VM and maintaining a data structure containing VM, ID, job ID, and VM status [15]. Round robin load balancer is an algorithm that migrates jobs from an overloaded VM to an underutilized VM based on the least hop time. Weighted round robin is a modified version of Round Robin, assigning a relative weight to all VMs [16]. It increases the number of jobs serviced by cloud providers, improving working and business performance.

Throttled load This algorithm ensures only a pre-defined Response time improved but other balancer number of internet cloudlets are allocated to a parameters are not taken into account single VM at any given time. such as: weight of VM, processing time, etc.

Modified throttled Focuses mainly on how incoming jobs are Resource utilization response time has algorithm assigned to the available VMs. improved.

Load nearly distributed uniformly among VMs.

Efficient VM load The proposed algorithm finds the expected Increases performance of the cloud balancing algorithm response time of each resource. environment. Decreases response time and cost.

Active VM load maintains information about each VM and the Does not consider the hardware balancer number of requests currently allocated to the capacity of VMs.

VMs WALB Allocates weighted count according to the Increase response time and Algorithm computing power of the VM, but the processing time algorithm does not consider process duration for each individual request.

Modified WALB This algorithm identifies VM with least load, Processing time: hence they tried best Algorithm least process duration and most powerful to consider the most affecting factor VM according to the weight assigned but it (process duration) in performance considers process duration. increase.

Enhanced load Propose a technique to avoid deadlock among Improves: migration time performance balancing to avoid VMs while processing a request by response time deadlock migrating the VM.

Round robin load Data center controller assigns first request to There is a possibility that some nodes balancer a VM, picked randomly from the group. It may get heavily loaded while others are assigns requests to the rest of VMs in overloaded.

circular order. Decrease resource utilization

Weighted round This algorithm assigns a relative weight to all Improvement of resource utilization robin algorithm the VMs.

Dynamic load When the users send the request to the Fault tolerance high scalability low balancing: improve dynamic load balancer, it gathers the processor overhead efficiency in cloud utilization and memory utilization of each Computing active server.

Round robin with The limitation of round robin algorithm is that Improved response time data center server affinity: a VM it does not save the state of the previous processing time Load balancing allocation of a VM to a request while the algorithm for cloud same state is saved in the proposed algorithm.

based infrastructure

The dynamic load balancer monitors the load of each VM in the cloud pool. If the processor and memory utilization are less than 80%, it instantiates a new VM on the following server with the lowest processor and memory utilization. The algorithm also checks the fault occurrence of a server, and if any fault occurs, the VMs will be shifted to another server with less than 80%. The proposed algorithm achieves high scalability, dynamic load balancing, fault tolerance and low overhead. The modified throttled algorithm [16] attempts to improve the response time and efficiency of load balancing in cloud computing. It uses a VM state list to store each VMs allocation status (i.e., Busy/Available) and an index table of VMs and the state of VMs. The VM at first index is initially selected depending upon the state of the VM, and the VM at index next to the already assigned VM is chosen depending on the state of the VM When compared to existing round-robin and throttled algorithms, the response time for proposed algorithm has improved considerably.

Enhanced load balancing algorithm using efficient cloud management system is based on the least hope time, while WALB identifies the least loaded and most powerful VM In round robin LB, the request is assigned circularly, but a new algorithm, "weighted round robin", is proposed to improve load balancing and response time. Dynamic load balancer aims to improve efficiency in cloud computing by monitoring the load of each VM and saving the previous state of allocation of a VM to a request from a user. The modified throttled algorithm aims to improve response time and efficiency.

In round robin load balancer [17], [18] the data center controller randomly chooses a VM from the group to receive the initial request. It then distributes requests to the VMs in a circular fashion. A VM gets moved to the end of the list once a request has been assigned. The round robin algorithm's advantage is that it eliminates the need for inter-process communication. As the duration of each process cannot be predicted before it is performed, it is possible that some nodes could become overloaded.

Cloud computing provides everything to the user as a service, including application, platform, and infrastructure. Load balancing is required to distribute the load evenly among all servers in the cloud to maximize resource utilization, increase throughput, provide good response time, and reduce energy consumption. The threshold algorithm guarantees most of these metrics, except for overload rejection, fault tolerant, and process migration. We plan to improve throttled to make it more suitable for the cloud environment and more efficient regarding process migration.

Conceptualization, R. G. K and Sh. S; methodology, R. G. K; software, F. R; validation, R. G.K. and F. R; formal analysis, Sh. S; investigation; writing—original draft preparation, R. G. K; writing—review and editing, Sh. S. All authors have read and agreed to the published version of the manuscript.

This research received no external funding.

All the data are available in this paper.

The authors declare no conflict of interest.