Research Article

An In-depth Benchmarking of Evolutionary and Swarm Intelligence Algorithms for Autoscaling Parameter Sweep Applications on Public Clouds

Table 14

Relevant related works summary.

Application typePaperAlgorithmAutoscalingPricing modelMetricsNumber of applications

BoT[12]NSGA-IIYesOn-demand, spotMakespan, cost, and OOB errors2
[14]NSGA-IIIYesOn-demand, spotMakespan, cost, and OOB errors2
[46]PSO - NNNoOn-demand, spotMakespan and cost1
[47]DEANoOn-demandMakespan and load balancing1
[42]Check-pointingNoOn-demand, spotCost1
[43]DDSNoOn-demandCost3
[48]NSGA-IIIYesOn-demandMakespan and cost2

Web[49]Cost-efficient and fault tolerantYesOn-demand, spotAvailability, cost, and RT1
[50]RLPASYesOn-demandCPU utilization, RT, and throughput3
[45]RHASYesOn-demandCost, RT, and QoS2
[51]ML based on reactive and proactiveYesOn-demandBroker profit and cost1
[52]NN—LRYesOn-demandRT and cost3

Workflows[53]SIAAYesOn-demand, spotMakespan, cost, and task failures4
[54]Online cost-efficient schedulingNoOn-demand, spotCost4
[55]Dynamic approachNoOn-demand, spotCost, reliability, and fault tolerance1
[41]Dynamic autoscaling based on EDFYesOn-demandCost3
[56]Scaling firstYesOn-demandCost3
[57]Dynamic approachYesOn-demandMakespan and cost1
[58]NSGA-IIYesOn-demand and spotMakespan, cost, and OOB errors4