Attack recovery from malicious transactions in distributed database systems

Abstract

Database protection mechanisms often fail to prevent occurrence of malicious transactions. In this paper, we consider the problem of database recovery from such committed malicious transactions in distributed database systems. In a database, the result of one transaction may affect the execution of some of the later transactions. This leads to damage spreading, which makes attack recovery even more complex. Traditional recovery schemes usually perform complete rollback to undo the effect of all the transactions, both malicious as well as non-malicious. We define several useful dependency relationships among transactions and present an algorithm to restore the consistency of a distributed database by negating the effect of only those transactions that are directly or transitively dependent on the malicious transactions.