应用随机Petri网分析分布式信息系统可生存性  

Analyzing Survivability of Distributed Information System using SPN

作　　者：孙显军[1,2] 朱亮[3] 高志民[4] 林闯[1] 刘卫东[1] 

机构地区：[1]清华大学计算机系 [2]中国人民解放军61741部队 [3]中国铝业股份有限公司信息部 [4]北京交通大学

出　　处：《系统仿真学报》2008年第S2期181-186,193,共7页Journal of System Simulation

基　　金：中国博士后基金(20080430040)

摘　　要：现代社会高度依赖于各种分布式信息系统。因此,当出现攻击、网络失效和自然灾害等意外事件时,重要服务必须保证可用。针对分布式信息系统提出一种基于随机Petri网的可生存性模型分析方法,评价其关键服务失效情况下的系统可生存性。为避免随机Petri网求解大规模系统面临的状态爆炸问题,首先对模型进行化简,然后建立结合性能和可用性的双层模型,分析不同失效和修复速率对性能、可维护性、可靠性和可用性的影响。结合一个典型的分布式信息系统实例——中国网上招生录取系统,给出了详细的建模和求解过程,最后通过Matlab仿真,直观地比较出可生存性中的测量指标随着失效和修复参数的变化情况。Our society is heavily dependent on a wide variety of distributed information system.The essential service of distributed information system must be available even when undesirable events such as attack,network failures,or natural disasters happen.In this paper we assess the survivability of distributed information with essential service exposed to failures.We have developed analytic models using stochastic Petri net.In order to avoid state space explosion while addressing large networks we reduce our models firstly,then give a composite model by combining performance and recovery models.The modeling approaches are applied to an information system example.Finally,the simulation by Matlab shows us the measures of survivability varying with repair rate and failure rate clearly.

关 键 词：可生存性 分布式信息系统 随机PETRI网 中国网上招生系统 

分 类 号：TP393.09[自动化与计算机技术—计算机应用技术]