APPROACH TO INTRUSION TOLERANCEMETHOD FOR LOG SERVICE SYSTEM BASED ON BAYESIAN BELIEF NETWORK

摘要

In order to improve survivability of an application system,a method is represented about how to protect log records stored in log servers and how to provide recovery information after an application system is intruded,then an intrusion tolerance strategy is introduced that a log record is divided into different portions,these portions are stored into many different log servers,and can be combined together to form a correct log record when necessary.A Bayesian belief network is created,which can be used to judge an anomaly access and its request type according to the feature data from request information of a user so as to locate the attacked log servers and the destroyed portions quickly in massive information and recover a log service system at a lowest cost.This method can protect veracity and integrality of log records to some extent.