Font Size: a A A

Research On Abnormal Access Request Detection Based Secure Subsystem Of Storage Area Network System

Posted on:2009-02-11Degree:DoctorType:Dissertation
Country:ChinaCandidate:T CaiFull Text:PDF
GTID:1118360275451125Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
The storage area network system is the hot topic in current researching.It can provide the shared storage service with very large storga space.And it is used in many information system as the data center.So its security is very important problem.The storage area network system is mixture of storage and network.The large time and space consumption need when using current secure technology to ensure large data stored in storage area network system.Then the storage area network system will lose many I/O performance.So researching the scure sub-system for storage area network system with low and stabile time and space consumption is very urgent problem.The artificial immune algorithm can discern the abnormal information efficiently. We use it to research and design the secure sub-system based on abnormal access request detection,and realize the secure storage area network system with high I/O performance and high security.Because the metadata server and intdligent disk can work in parallel,we add the central abnormal access request detection module in the metadata server and add the associate abnormal access request detection module in intelligent disk.It can avoid performance bottleneck of secure sub-system.We present active access request checking strategy and design the distributed structure of secure sub-system.It can reduce the time and space consumption of protecting the storage area network system.We define the detector-number to instead the binary string detector,and present the constraint-based detector-number generating algorithm.It can solve some problems in current detector generating algorithm and access request checking algorithm such as: the too large time and space consumption,redundant consumption and difficuility analyzing the accuracy of checking out abnormal access request.Basing on the composition of access request,the numeric-mixed matching rule is presented to improve the accuracy and reduce the time and space consumption of checking access request.Then the prototype of the central and associate abnormal access request detection module are realized,and the testing result shows it can detect the abnormal access request with low time and space consumption.The selfadaptation threshold selection algorithm and detector-number set rapid changing algorithm are presented.They can improve the adaptation of abnormal access request detection module.Designing the access request checking algorithm and detecting abnormal access request in storage area network system.The detector-number distribution algorithm is givened to distribute the detector-number between central abnormal access request detection module and associate abnormal access request detection module.Analyzing the relation between the metadata server and intelligent disk,the game-based optimization algorithm is presented.The game model between central abnormal access request detection module and associate abnormal access request detection module is builded and the utility function is givened,to optimize the time and space consumption of secure sub-system.The open simulator of storage area network system is realized to simulate the process of dealing with access request in storage area network system.Then the prototype of secure sub=system based on abnormal access request detection is realized. Using the real trace to testing,the result shows the secure sub-system can detect the abnormal access request with high accuracy,and time and space consumption of cheching access request is low and stable.On the open souce storage area network system named Lustre,the secure sub-system based on abnormal access request detection is realized.It used to setup the prototype of secure storage area network system.The Iozone is used to test the I/O performance of it and compare with Lustre.The result proves the loss of the I/O performance is low,it is 5%to 10%.
Keywords/Search Tags:storage area network, secure storage, artificial immune algorithm, game theory
PDF Full Text Request
Related items