A Comparative Analysis of NSGA-II and NSGA-III for Autoscaling Parameter Sweep Experiments in the Cloud

Table 5

Results obtained from the computational experiments. The rows present the results per autoscaler regarding all the studied applications and sizes. For the average makespan, cost, number of task failures, and L₂-norm, lower values represent better results.

Application	Size	Autoscaler	Makespan	Cost	Task failures	L2-norm
Plate3D	30	NSGA-III	1142.11	0.06	5.80	0.26
		NSGA-II	1381.02	0.23	4.53	0.44
	100	NSGA-III	1910.03	0.12	12.05	0.17
		NSGA-II	1836.50	0.36	25.77	0.28
	300	NSGA-III	2894.53	0.39	62.32	0.23
		NSGA-II	2720.12	0.98	82.17	0.37
Cruciform	30	NSGA-III	3011.33	0.26	44.75	0.30
		NSGA-II	3599.30	0.47	15.80	0.42
	100	NSGA-III	2925.29	0.77	146.33	0.29
		NSGA-II	3602.92	0.91	59.92	0.46
	300	NSGA-III	4223.02	2.22	309.80	0.18
		NSGA-II	5118.21	4.74	219.83	0.30
Ensemble	60	NSGA-III	5095.71	0.26	35.32	0.30
		NSGA-II	5504.60	0.69	23.33	0.46
	200	NSGA-III	5868.27	0.91	147.00	0.34
		NSGA-II	6912.26	1.78	118.07	0.54
	600	NSGA-III	5840.39	2.80	372.86	0.31
		NSGA-II	7401.15	4.59	281.16	0.40