An Extended Transaction Model Approach for Multilevel Secure Transaction Processing
Vijayalakshmi Atluri, Ravi Mukkamala
Abstract
Open-access reader
Vijayalakshmi Atluri, Ravi Mukkamala
Abstract
Open-access reader
Multilevel secure transaction processing has been well explored in the past decade. Despite this research, the proposed secure concurrency control protocols are not completely satisfactory because of the stringent constraints imposed by multilevel security. In this paper, we argue that modeling a transaction as an extended transaction model could significantly reduce the performance penalty. We accomplish this by minimizing (1) the probability of restarting a high security level transaction; and (2) the portion of the transaction to be reexecuted, when a restart becomes inevitable. In particular, we exploit the non-flat nature of transactions by identifying dependencies among various components of a transaction and portraying a transaction as an advanced transaction model. We demonstrate, via formal proofs, that our approach preserves the semantics of the transaction, and our concurrency control algorithm guarantees serializability.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Multilevel secure transaction processing has been well explored in the past decade. Despite this research, the proposed secure concurrency control protocols are not completely satisfactory because of the stringent constraints imposed by multilevel security. In this paper, we argue that modeling a transaction as an extended transaction model could significantly reduce the performance penalty. We accomplish this by minimizing (1) the probability of restarting a high security level transaction; and (2) the portion of the transaction to be reexecuted, when a restart becomes inevitable. In particular, we exploit the non-flat nature of transactions by identifying dependencies among various components of a transaction and portraying a transaction as an advanced transaction model. We demonstrate, via formal proofs, that our approach preserves the semantics of the transaction, and our concurrency control algorithm guarantees serializability.
Key concepts: Serializability, Computer science, Distributed transaction, Transaction processing system, Database transaction, Transaction processing, Online transaction processing, Distributed computing