Load balancing in cloud computing – A hierarchical taxonomical classification

Afzal, Shahbaz; Kavitha, G.

doi:10.1186/s13677-019-0146-7

Journal of Cloud Computing

Advances, Systems and Applications

Table 1 Proactive approaches of Cloud load balancing in existing literature

From: Load balancing in cloud computing – A hierarchical taxonomical classification

Reference	Algorithm Used	Trait Used	Type of Load Balancing	Technique involved	Algorithm Complexity	Advantages	Disadvantages
[40]	Conventional Non Classical	Task Scheduling	Task LB	Heuristic (Classical, Deterministic)	Not Specified	Capable of handling heavy workloads within predefined deadline.	Tasks whose execution time is more than defined deadline are rejected.
						Provides enhanced elasticity.	Thresholds for defining overloaded and under loaded VMs are set arbitrarily without formulating equation for them.
						Minimize makespan with improved task acceptance ratio.
						Minimize task rejection ratio	The experimental are run on Cloudsim using space shared policy only and not time shared policy.
						Perform automatic scaling of resources
[41]	Full Set algorithm and Column generation algorithm	VM scheduling	VM LB	Optimization (Classical, Deterministic, LP)	[O(2)^N – n O(2^k)/2]	Load balancing is performed among minimum number of VMs	Algorithm evaluates only single objective function.
						Improved resource utilization	The experiments are run on C++ programs
						Resource over provisioning is avoided
						The algorithm runs in real-time scale with simple complexity.
[37]	Dragonfly optimization and constraint measure-based load balancing	Task Scheduling	Task LB	Optimization (Swarm Based)	Not Specified	Load balancing is performed with less power consumption	Cannot handle tasks beyond threshold limit.
[37]		Task Scheduling	Task LB	Optimization (Swarm Based)	Not Specified	Load balancing is performed with less power consumption	Task rejection ratio is high
[38]	Fairness Aware Algorithm	Resource Scheduling	CPU LB	Optimization (non cooperative game theory based)	Not Specified	Optimal Lb is achieved at Nash equilibrium point.	High task execution time
[38]	Fairness Aware Algorithm	Resource Scheduling	CPU LB	Optimization (non cooperative game theory based)	Not Specified	Minimize expected response time	High task execution time
[42]	Honey Bee Behaviour	Task Scheduling	Task LB	Optimization (Swarm Based)	Not Specified	Low response time.	Low scalability
[42]	Honey Bee Behaviour	Task Scheduling	Task LB	Optimization (Swarm Based)	Not Specified	Low makespan	Low scalability
[43]	ACO	Task Scheduling	Task LB	Optimization (Swarm Based)	Not Specified	Less makespan	Tasks are mutually independent
[43]	ACO	Task Scheduling	Task LB	Optimization (Swarm Based)	Not Specified	Measures degree of imbalance among VMs	Memory intensive tasks are not taken
[44]	Agent based Nature Inspired Algorithm	Resource Scheduling	Resource LB	Metaheuristic	Not Specified	High scalability	Execution cost not considered
						Less response time	Service level violations not considered
						Improved resource utilization	Task rejection rate not considered
[45]	Non- Classical	Resource Scheduling	Resource LB	Heuristic	Not Specified	High fault tolerance	High response time
						Less overhead	High execution time
						Less overhead	High makespan
[46]	Weighted Round Robin	Resource Scheduling	Server LB	Heuristic	O(1)	Good resource utilization	response time not chosen
						Enhanced throughput	degree of balance not chosen
						Less overhead	energy efficiency not chosen
						High fault tolerance	energy efficiency not chosen
[47]	Nature Inspired GA	Task and Resource Scheduling	Task LB	Optimization	O(1)	Efficient resource utilization	Priority based
						Less resource wastage	Less scalability
						Small energy consumption	Less fault tolerance
						Less SLV
						Improved degree of balance

Back to article page