基于检查点的分布式软件监控与可信性评价

doi:10.11959/j.issn.1000-436x.2016048

摘要/Abstract

摘要：

基于检查点的传统软件可信性评价方法以及目前针对分布式软件的交互关联规则，对于具有复杂交互行为的分布式软件均不适用。采用伴随式分布式软件监控机制，在节点内织入3类检查点，引入适应复杂交互场景的交互关联规则。通过将节点分解为多层模块结构，进行基于检查点结构树的节点实例可信性以及基于节点的分布式软件可信性评价。实验表明能以较小的监控开销，更准确地评价分布式软件实例的可信性，能够处理无限路径，且不存在大状态空间问题。

关键词: 分布式软件, 可信性, 检查点, 交互, 行为监控

Abstract:

Traditional software trustworthiness evaluation based on checkpoints and the existing interaction association rules for distributed software were inapplicable to distributed software with complex interactions.According to this problem,an accompanying distributed software monitoring mechanism was used,and three types of checkpoints in nodes and interaction association rules were introduced for complex interaction scene.The trustworthiness of the node instance was evaluated based on checkpoint structure tree by dividing nodes into several modules,and then the trustworthiness of distributed soft-ware was evaluated based on nodes.The experimental results showed that the approaches could evaluate the trustworthiness of distributed software instance accurately with small monitoring cost,and could be suitable for infinitely long paths with no state explosion problem for trustworthiness evaluation of distributed software.

Key words: distributed software, trustworthiness, checkpoint, interaction, behavior monitoring

李珍,田俊峰,常卓,马晓雪. 基于检查点的分布式软件监控与可信性评价[J]. 通信学报, 2016, 37(3): 7-19.

Zhen LI,Jun-feng TIAN,Zhuo CHANG,Xiao-xue MA. Distributed software monitoring and trustworthiness evaluation based on checkpoints[J]. Journal on Communications, 2016, 37(3): 7-19.

图/表 15

图1

图2

图3

图4

图5

图6

图7

图8

图9

图10

图11

图12

表1

表2

表3

参考文献 19

[1]	REYNOLDS P , KILLIAN C , WIENER J L , et al. Pip:detecting the unexpected in distributed systems[C]// The 3rd Symposium on Net-worked Systems Design and Implementation. USENIX Assoc ion Berkeley,CA,USA, c2006:115-128.
[2]	杨晓晖，周学海，田俊峰，等. 一个新的软件行为动态可信评测模型[J]. 小型微型计算机系统， 2010,31(11): 2113-2120. YANG X H , ZHOU X H , TIAN J F , et al. Novel dynamic trusted evaluation model of software behavior[J]. Journal of inese Computer Systems, 2010,31(11): 2113-2120.
[3]	刘玉玲，杜瑞忠，冯建磊，等. 基于软件行为的检查点风险评估信任模型[J]. 西安电子科技大学学报(自然科学版)， 2012,39(1): 179-184. LIU Y L , DU R Z , FENG J L , et al. Trust model of software behaviors based on check point risk assessment[J]. Journal of Xidian University, 2012,39(1): 179-184.
[4]	田俊峰，张亚姣 . 基于马尔可夫的检查点可信评估方法[J]. 通信学报， 2015,36(1): 230-236. TIAN J F , ZHANG Y J . Checkpoint trust evaluation method based on Markov[J]. Journal on Communications, 2015,36(1): 230-236.
[5]	刘东红，郭长国，王怀民，等. 监控使能的分布式软件系统构造方法[J]. 软件学报， 2011,22(11): 2610-2624. LIU D H , GUO C G , WANG H M , et al. Monitoring enabled istri-buted software construction method[J]. Journal of Software, 2011,22(11): 2610-2624.
[6]	文志诚，李长云，满君丰 . 基于贝叶斯网的分布式软件行为运行时可信性分析[J]. 小型微型计算机系统， 2012,33(3): 504-511. WEN Z C , LI C Y , MAN J F . Analyzing running-time behavioral cre-dibility for distributed software based on Bayesian network[J]. Journal of Chinese Computer Systems, 2012,33(3): 504-511.
[7]	彭成，杨路明，满君丰 . 网络化软件交互行为动态建模[J]. 电子学报， 2013,41(2): 314-320. PENG C , YANG L M , MAN J F . Dynamic modeling of networked software interactive behavior[J]. Acta Electronica Sin, 2013,41(2): 314-320.
[8]	WANG Q , LU Y , XU Z J , et al. Software reliability mode for com-ponent interaction mode[J]. Journal of Electronics(Chi 2011,28(4/5/6): 632-642.
[9]	YANG N H , YU H Q , QIAN Z L , et al. Modeling and quantitatively predicting software security based on stochastic Petri nets[J]. Mathe-matical and Computer Modeling, 2012,55(1-2):102-112.
[10]	TYAGI K , SHARMA A . An adaptive neuro fuzzy model for estimat-ing the reliability of component-based software systems[J]. Applied Computing and Informatics, 2014,10(s1-2):38-51.
[11]	CHEN X H , LIU J , LIU Z M . Requirements monitoring for nternet-ware:an interaction based approach[J]. Science China:Information Science, 2013,56(8):1-5.
[12]	GOKHALE S S . Architecture-based software reliability analysis:overview and limitations[J]. IEEE Transactions on Dependable and Secure Computing, 2007,4(1):32-40.
[13]	CHEUNG R C . A user-oriented software reliability model[J]. IEEE Transactions on Software Engineering, 1980,SE-6(2):118-125.
[14]	GOKHALE S S , WONG W E , HORGAN J R , et al. An analytical approach to architecture-based software performance and reliability prediction[J]. Performance Evaluation, 2004,58(4):391-412.
[15]	HSU C J , HUANG C Y . An adaptive reliability analysis using path testing for complex component-based software systems[J]. IEEE Transactions on Reliability, 2011,60(1):158-170.
[16]	YACOUB S , CUKIC B , AMMAR H H . A scenario-based reliability analysis approach for component-based software[J]. IEEE Transac-tions Reliability, 2004,54(4):465-480.
[17]	李珍，田俊峰，赵鹏远 . 基于分级属性的软件监控点可信行为模型[J]. 电子与信息学报， 2012,34(6): 1445-1451. LI Z , TIAN J F , ZHAO P Y . A trustworthy behavior model for software monitoring point based on classification attributes[J]. Journal of Electronics ＆ Information Technology, 2012,34(6): 1445-1451.
[18]	李珍，田俊峰，杨晓晖 . 基于检查点分级属性的软件动态可信评测模型[J]. 计算机研究与发展， 2013,50(11): 2397-2405. LI Z , TIAN J F , YANG X H . Dynamic trustworthiness eval ion model of software based on checkpoint's classification attributes[J]. Journal of Computer Research and Development, 2013,50(11): 2397-2405.
[19]	邵建平，王玲 . “拉长板”木桶原理及其运用研究[J]. 科技管理研究， 2005,25(4): 159-161. SHAO J P , WANG L . Research on “stretched panels” bucket theory and its application[J]. Science and Technology Manage nt Research, 2005,25(4): 159-161.

方法	交互关联类型	符号	关联规则	交互关联可信性值
	因果	→	placeOrder→writeToDB,writeToDB→notifySupplier ,	交互关联可信性值
文献[7]	并发	\|\|	changeOrder→writeToDB,notifyUserOfShipment→writeToDB ,	1
	重复	?	writeToDB→notifyUser,changeOrder→noChangeAllowed
			placeOrder ＜ _pwriteToDB ＜ _pnotifySupplier,
文献[11]	因果	＜_p	changeOrder ＜ _pwriteToDB ＜ _pnotifySupplier,notifyUserOfShipment ＜ _pwriteToDB ＜ _pnotifyUser,	1
			changeOrder ＜ _pnoChangeAllowed	0
本文方法	同步因果异步因果并发互斥重复	＜_S ＜_A\|\|+*	placeOrder ＜_SwriteToDB ＜_SnotifySupplier,[change order before shipment](changeOrder ＜_S writeToDB ＜_S notifySupplier),notifyUserOfShipment ＜_SwriteToDB ＜_SnotifyUser,[change order after shipment](changeOrder ＜_S noChangeAllowed)	0

方法	容纳全部节点实例	处理无限路径	处理大状态空间	适用场合
方法	容纳全部节点实例	处理无限路径	处理大状态空间	节点实例可信性评价	节点可信性评价
基于路径^[15,16]	×	×	v	v	×
基于状态^[13,14]	v	v	×	×	v
基于检查点结构树	v	v	v	v	v

测试序号	2)writeToDB		3)notifySupplier		8)writeToDB		9)notifyUser		场景 1
测试序号	t_get	t_exe	t_get	t_exe	t_get	t_exe	t_get	t_exe	场景 1
1	10.35	15.74	10.05	8.25	10.23	15.87	8.98	7.65	1 124.34
2	9.98	14.95	9.30	7.83	10.11	15.44	9.24	7.87	1 098.35
3	9.57	14.80	9.91	8.13	9.76	14.32	8.65	7.54	1 087.65
4	10.12	15.56	9.54	7.95	9.99	15.34	8.50	7.39	1 128.72
5	9.76	14.55	9.58	8.01	9.74	14.78	8.78	7.78	1 105.60
平均值	9.96	15.12	9.68	8.03	9.97	15.15	8.83	7.65	1 108.93
额外开销均值	0		1.65		0		1.18		2.83