Coarse-Grain QoS-Aware Dynamic Instance Provisioning for Interactive Workload in the Cloud

Table 3

Parameters for two function approximation approaches.

Decision epoch	Linear approximation		Nonlinear approximation
	𝑎	𝑏	𝑎	𝑏
1	0.910558	−0.03026	1.96𝐸+13	−10.3672
2	1.17268	−0.03832	3.42𝐸+13	−10.4338
3	1.31243	−0.04182	6.23𝐸+13	−10.5444
4	1.50093	−0.04697	4.04𝐸+13	−10.3147
5	1.46158	−0.04394	1.62𝐸+13	−9.97221
6	1.59491	−0.04727	3.68𝐸+12	−9.43435
7	1.49158	−0.04142	5.56𝐸+11	−8.76771
8	1.57272	−0.043	2.35𝐸+11	−8.44603
9	1.63012	−0.04319	1.16𝐸+11	−8.13732
10	1.6601	−0.04237	5.02𝐸+10	−7.84783
11	1.75972	−0.04396	1.81𝐸+10	−7.47832
12	1.81333	−0.04415	6.70𝐸+09	−7.12764
13	1.81285	−0.04386	8.02𝐸+09	−7.17651
14	1.75933	−0.0437	1.76𝐸+10	−7.4644
15	1.7186	−0.04436	4.01𝐸+10	−7.77709
16	1.63952	−0.04352	6.46𝐸+10	−7.9815
17	1.58403	−0.04322	1.83𝐸+11	−8.36223
18	1.55035	−0.04373	1.20𝐸+12	−9.01639
19	1.54785	−0.0454	3.54𝐸+12	−9.41871
20	1.38708	−0.04119	5.03𝐸+12	−9.60535
21	1.50875	−0.04732	8.56𝐸+13	−10.5575
22	1.31375	−0.04188	2.37𝐸+13	−10.2367
23	1.16168	−0.03787	3.70𝐸+13	−10.4633
24	0.866371	−0.02836	1.52𝐸+13	−10.2863