Research On Data Isolation Base On Ioslation Classification

As one of the key parts of the information system, a database collectively stores much important and confidential information. The development of the network makes databases primary targets for hackers. Once a database is attacked, the society and economy will suffer huge loss. Damaged data isolation technology can let the attacked database function in a normal way while prevent the damaged data from spreading. Such a technology is now becoming more and more important since traditional databases have no effective protection against vicious attack.Comparing several traditional isolation technologies and analyzing the requirements on the function and quality of damaged data isolation system, we designed a system based on the graded damaged data isolation system to improve the safety of general databases. The designed system was made up of two modules: damaged data management module and transaction filter module. Damaged data management module primarily manages the damaged data and provide information about the damaged data for the transaction filter module. Bitmap method and Mixed bitmap partition list method to manage damaged data. Transaction filter module checks transactions on three isolation grades and isolates damaged data in different ways to prevent their spreading. The alert level will give warning information when the system detected damaged data to manipulated by transaction; Filter level allow transaction visit normal data; however, the refusing level will refuse the transaction. The isolation granularity used in this thesis is dynamic adjustment granularity according to certain criteria.Results of performance test show that the graded damaged data isolation system has less damage for the whole system and our design goals have been obtained.