云化虚拟化网络业务可用度多参数敏感性分析

doi:10.12305/j.issn.1001-506X.2022.08.34

摘要/Abstract

摘要：

网络功能的虚拟化使得网络中软硬件设备类型变多, 不同的设备有不同的故障修复参数, 这些参数均会对业务可用度造成影响。由于不同的业务共享网络资源难以找到对业务可用度影响最大的关键设备参数，现有的局部敏感性分析方法难以保证结果的正确性。全局敏感性分析方法需要大量的仿真样本，操作困难，所以本文将网络中的参数分为异类和同类, 对异类参数采用全效应指数法，对同类参数采用动态业务介数法, 最终综合两种方法分析的结果找到网络中的关键参数。该方法能够保证结果准确性的同时极大减少仿真的样本量。

关键词: 云化虚拟化网络, 业务可用度, 敏感性分析, 关键参数

Abstract:

Hardware and software device types are increasing with the virtualization of network functions. Different devices have different fault and recovery parameters, which affect application availability. It is difficult to find the key parameters that have the greatest impact on application availability, because different applications share network device resources. The correctness of the results are difficult to be ensured by existing local sensitivity analysis methods, and existing global sensitivity analysis methods require a large number of simulation samples and are difficult to be conducted. Therefore, this paper divides the parameters in the network into heterogeneous and homogeneous ones, and then analyses them by the total effect index method and the dynamic application intermediate method respectively. Finally, the key parameters in the network are found by synthesizing the analysis results of the two methods. The proposed method can ensure the accuracy of the results and greatly reduce the sample size of simulation.

Key words: network function virtualization (NFV), application availability, sensitivity analysis, key parameters

中图分类号:

N94-0

朱杰, 黄宁, 程亮. 云化虚拟化网络业务可用度多参数敏感性分析[J]. 系统工程与电子技术, 2022, 44(8): 2677-2687.

Jie ZHU, Ning HUANG, Liang CHENG. Multi-parameter sensitivity analysis of network function virtualization application availability[J]. Systems Engineering and Electronics, 2022, 44(8): 2677-2687.

图/表 14

表1

图1

图2

表2

图3

表3

表4

表5

表6

表7

不同样本容量下的异类参数敏感性排序情况"

参数类别	样本容量n
参数类别	20	20	40	40	60	60	80	80	均值
x₁	0.016 4	0.012 4	0.015 8	0.010 9	0.011 7	0.008 6	0.018 7	0.018 0	0.014 0
x₂	0.009 9	0.010 0	0.035 1	0.018 9	0.012 5	0.016 1	0.017 0	0.017 3	0.017 1
x₃	0.001 8	0.001 8	0.005 2	0.004 8	0.002 4	0.002 3	0.004 6	0.003 5	0.003 3
x₄	0.356 0	0.332 1	0.499 6	0.386 0	0.370 0	0.387 9	0.495 2	0.781 6	0.451 0
x₅	0.003 2	0.002 7	0.008 1	0.006 3	0.002 5	0.003 9	0.005 6	0.004 4	0.004 5
x₆	0.002 0	0.002 2	0.003 9	0.005 1	0.002 1	0.002 1	0.003 9	0.003 4	0.003 0
x₇	0.003 9	0.000 7	0.003 2	0.005 9	0.003 2	0.002 9	0.003 8	0.004 0	0.003 4
x₈	0.002 3	0.001 4	0.002 1	0.004 2	0.003 4	0.002 4	0.004 1	0.003 0	0.002 8
x₉	0.003 1	0.001 9	0.003 7	0.005 0	0.002 6	0.002 3	0.003 3	0.002 9	0.003 1
x₁₀	0.002 8	0.001 5	0.004 5	0.004 0	0.003 6	0.002 7	0.003 5	0.003 7	0.003 2
x₁₁	0.001 7	0.000 8	0.002 7	0.005 3	0.003 9	0.002 3	0.003 4	0.005 1	0.003 1
x₁₂	0.001 6	0.001 6	0.002 8	0.004 2	0.003 0	0.001 9	0.003 6	0.003 7	0.002 8
x₁₃	0.000 8	0.001 3	0.003 8	0.005 2	0.003 3	0.002 8	0.003 0	0.004 9	0.003 1
x₁₄	0.003 6	0.001 3	0.006 4	0.005 0	0.003 3	0.002 7	0.004 4	0.005 0	0.003 9
x₁₅	0.001 4	0.003 1	0.004 2	0.004 1	0.003 5	0.002 7	0.005 3	0.002 8	0.003 3
x₁₆	0.337 0	0.382 3	0.567 9	0.570 0	0.580 4	0.195 2	0.268 2	0.853 7	0.469 3
x₁₇	0.001 5	0.004 1	0.005 2	0.003 4	0.003 2	0.002 5	0.005 1	0.005 1	0.003 7
x₁₈	0.002 0	0.002 2	0.007 1	0.008 1	0.003 4	0.004 5	0.003 6	0.006 6	0.004 6
x₁₉	0.001 3	0.001 1	0.004 6	0.009 1	0.003 2	0.003 1	0.005 8	0.003 3	0.003 9
x₂₀	0.004 4	0.006 3	0.007 6	0.014 5	0.007 5	0.006 0	0.009 4	0.009 9	0.008 2
x₂₁	0.006 1	0.003 2	0.014 9	0.012 4	0.008 0	0.007 0	0.017 0	0.012 8	0.010 1
x₂₂	0.003 5	0.001 4	0.003 2	0.008 3	0.002 8	0.002 5	0.003 4	0.004 2	0.003 6
x₂₃	0.225 2	0.126 0	0.324 9	0.376 1	0.492 7	0.216 9	0.349 4	0.427 8	0.317 3
x₂₄	0.006 1	0.003 9	0.005 3	0.007 4	0.003 6	0.002 7	0.006 3	0.003 9	0.004 9
x₂₅	0.003 8	0.002 4	0.003 8	0.004 3	0.002 7	0.002 3	0.003 2	0.004 0	0.003 3

表7

表8

表9

图4

表10

参考文献 18

1	黄宁. 网络可靠性及评估技术[M]. 北京: 国防工业出版社, 2020.
	HUANG N . Network reliability and evaluation technology[M]. Beijing: National Defense Industry Press, 2020.
2	YI Z W, HUANG N, CAI H S T. Evaluating the application availability of intelligent optical networks based on the network evolution model[C]//Proc. of the 31th European Safety and Reliability Conference (ESREL), 2021.
3	FREY H C , PATIL S R . Identification and review of sensitivity analysis methods[J]. Risk Analysis, 2010, 22 (3): 553- 578.
4	CARPIO F , JUKAN A . Improving reliability of service function chains with combined VNF migrations and replications[J]. Computer Research Repository, 2017,
5	RUI L L , CHEN X S , GAO Z P , et al. Petri net-based reliability assessment and migration optimization strategy of SFC[J]. IEEE Trans.on Network and Service Management, 2021, 18 (1): 167- 181. doi: 10.1109/TNSM.2020.3045705
6	STEINER M , BOURINET J M , LAHMER T . An adaptive sampling method for global sensitivity analysis based on least-squares support vector regression[J]. Reliability Engineering & System Safety, 2019, (183): 323- 340.
7	REN J , ZHANG W B , YANG J . Morris sensitivity analysis for hydrothermal coupling parameters of embankment dam: a case study[J]. Mathematical Problems in Engineering, 2019,
8	杨英杰, 于永利, 张柳, 等. 装备维修保障仿真系统灵敏度分析与参数优化[J]. 系统工程与电子技术, 2016, 38 (3): 575- 581.
	YANG Y J , YU Y L , ZHANG L , et al. Sensitivity analysis and parameters optimization for equipment maintenance support simu- lation system[J]. Systems Engineering and Electronics, 2016, 38 (3): 575- 581.
9	史妍妍, 孙志礼, 闫明. 响应面方法在可靠性灵敏度计算中的应用[J]. 东北大学学报(自然科学版), 2009, 30 (2): 270- 273.
	SHI Y Y , SUN Z L , YAN M . Application of response surface methodology in sensitivity calculation of reliability[J]. Journal of Northeastern University (Natural Science), 2009, 30 (2): 270- 273.
10	RODRIGUEZ-CAMINO E , AVISSAR R . Comparison of three land-surface schemes with the Fourier amplitude sensitivity test (FAST)[J]. Tellus A: Dynamic Meteorology and Oceanography, 2016, 3 (50): 313- 322.
11	李传奇, 崔佳伟, 马梦蝶, 等. 泵站动态规划优化模型的全局敏感性分析方法及应用[J]. 水电能源科学, 2019, 37 (5): 159- 163.
	LI C Q , CUI J W , MA M D , et al. Global sensitivity analysis of dynamic programming optimization model of pumping station and its application[J]. Water Resources and Power, 2019, 37 (5): 159- 163.
12	JIN R C , CHEN W , SUDJIANTO A . Analytical metamodel-based global sensitivity analysis and uncertainty propagation for robust design[J]. Journal of Quality Technology, 2006, 38 (4): 333- 348. doi: 10.1080/00224065.2006.11918622
13	韩晓阳, 孟相如, 康巧燕, 等. 可靠性与拓扑感知的服务功能链备份保护方法[J]. 系统工程与电子技术, 2021, 43 (7): 1961- 1970.
	HAN X Y , MENG X R , KANG Q Y , et al. Reliability and topo-logy aware service function chain backup protection method[J]. Systems Engineering and Electronics, 2021, 43 (7): 1961- 1970.
14	阳勇, 孟相如, 康巧燕, 等. 基于流量优化的可靠服务功能链部署方法[J]. 系统工程与电子技术, 2021, 43 (10): 3017- 3025. doi: 10.12305/j.issn.1001-506X.2021.10.38
	YANG Y , MENG X R , KANG Q Y , et al. Reliable service function chain deployment method based on traffic optimization[J]. Systems Engineering and Electronics, 2021, 43 (10): 3017- 3025. doi: 10.12305/j.issn.1001-506X.2021.10.38
15	杨旭, 肖子玉, 邵永平, 等. NFV可靠性探讨[J]. 电信科学, 2017, 33 (7): 136- 143.
	YANG X , XIAO Z Y , SHAO Y P , et al. Discussion on the reliability of NFV[J]. Telecommunications Science, 2017, 33 (7): 136- 143.
16	刘晓路, 陈英武, 荆显荣, 等. 优化拉丁方试验设计方法及其应用[J]. 国防科技大学学报, 2011, 33 (5): 73- 77. doi: 10.3969/j.issn.1001-2486.2011.05.015
	LIU X L , CHEN Y W , JING X R , et al. Optimized latin hypercube sampling method and its application[J]. Journal of National University of Defense Technology, 2011, 33 (5): 73- 77. doi: 10.3969/j.issn.1001-2486.2011.05.015
17	潘星, 张振宇, 张艳梅, 等. 基于Sobol敏感性分析的装备体系保障效能评估[J]. 系统工程与电子技术, 2021, 43 (2): 390- 398.
	PAN X , ZHANG Z Y , ZHANG Y M , et al. Equipment SoS support effectiveness evaluation based on Sobol sensitivity analy- sis[J]. Systems Engineering and Electronics, 2021, 43 (2): 390- 398.
18	崔明明, 汪鹏. 空空作战飞机敏感性分析研究[J]. 航空电子技术, 2020, 51 (3): 34- 40.
	CUI M M , WANG P . Sensitivity analysis of air-to-air combat aircraft[J]. Avionics Technology, 2020, 51 (3): 34- 40.

设备及参数类型	表示符号
数据中心出口网关(data center gateway, DCGW)	D
架顶交换机(top of rack, TOR)	T
行间交换机(end of row, EOR)	E
服务器(Server)	S
虚拟交换机(Vswitch)	Vs
虚拟机(virtual machine, VM)	V
平均故障间隔时间	α
故障检测率	β
故障检测时间	γ
自动维修率	δ
自动维修时间	ε
平均人工维修时间	ζ
故障倒换时间	η

参数类别	节点	全效应指数	节点动态业务介数	参数敏感度
T-γ	T1	μ1	e1	(μ1×e1)/(e1+e2+e3)
	T2	μ1	e2	(μ1×e2)/(e1+e2+e3)
	T3	μ1	e3	(μ1×e3)/(e1+e2+e3)
T-β	T1	μ2	e1	(μ2×e1)/(e1+e2+e3)
	T2	μ2	e2	(μ2×e2)/(e1+e2+e3)
	T3	μ2	e3	(μ2×e3)/(e1+e2+e3)
T-ε	T1	μ3	e1	(μ3×e1)/(e1+e2+e3)
	T2	μ3	e2	(μ3×e2)/(e1+e2+e3)
	T3	μ3	e3	(μ3×e3)/(e1+e2+e3)

节点类型	α/年	β	γ/s	δ	ε/min	ζ/h
DCGW	D-α	D-β	D-γ			D-ζ
EOR	T-α	T-β	T-γ			T-ζ
TOR	S-α	S-β	S-γ			S-ζ
Server	Vs-α	Vs-β	Vs-γ			Vs-ζ
Vswitch	V-α	V-β	V-γ	Vs-δ	Vs-ε	V-ζ
VM	D-α	D-β	D-γ	V-δ	V-ε	D-ζ

参数类别	符号表示	试验取值空间
D-α	x₁	[10, 100]
T-α	x₂	[10, 100]
S-α	x₃	[10, 100]
Vs-α	x₄	[1, 5]
V-α	x₅	[1, 5]
D-β	x₆	[0.5, 1]
T-β	x₇	[0.5, 1]
S-β	x₈	[0.5, 1]
Vs-β	x₉	[0.5, 1]
V-β	x₁₀	[0.5, 1]
D-γ	x₁₁	[10, 100]
T-γ	x₁₂	[10, 100]
S-γ	x₁₃	[10, 100]
Vs-γ	x₁₄	[100, 500]
V-γ	x₁₅	[100, 500]
Vs-δ	x₁₆	[0.5, 1]
V-δ	x17	[0.5, 1]
Vs-ε	x18	[2, 20]
V-ε	x19	[2, 20]
D-ζ	x20	[1, 10]
T-ζ	x21	[1, 10]
S-ζ	x22	[1, 10]
Vs-ζ	x23	[1, 10]
V-ζ	x24	[1, 10]
η/s	x25	[10, 100]

试验序号	A业务可用度	B业务可用度	C₁业务可用度
1	0.999 726 57	0.999 953 51	0.999 733 67
2	0.999 941 18	0.999 837 04	0.999 940 10
3	0.999 938 58	0.999 937 87	0.999 861 58
4	0.999 813 38	0.999 887 43	0.999 803 73
5	0.999 841 17	0.999 935 20	0.999 819 88
6	0.999 914 88	0.999 954 98	0.999 910 53
7	0.999 336 15	0.999 792 39	0.999 302 83
8	0.999 866 31	0.999 929 16	0.999 883 00
9	0.999 923 75	0.999 824 19	0.999 945 69
10	0.999 898 72	0.999 844 48	0.999 940 46
11	0.999 886 04	0.999 557 13	0.999 899 87
12	0.999 873 32	0.999 959 56	0.999 910 18
13	0.999 901 35	0.999 393 98	0.999 878 09
14	0.999 527 64	0.999 867 18	0.999 577 44
15	0.999 925 17	0.999 736 42	0.999 924 93
16	0.999 526 67	0.999 805 94	0.999 526 09
17	0.999 536 75	0.999 817 63	0.999 530 32
18	0.999 647 07	0.999 864 88	0.999 669 51
19	0.999 870 36	0.999 846 31	0.999 843 77
20	0.999 932 09	0.999 765 53	0.999 946 361