Secure Concurrency Control in Firm Real-Time Database Systems

  • Authors:
  • Binto George;Jayant R. Haritsa

  • Affiliations:
  • Database Systems Lab, Indian Institute of Science, Bangalore-560012, India. binto@dsl.serc.iisc.ernet.in;Database Systems Lab, Indian Institute of Science, Bangalore-560012, India. haritsa@dsl.serc.iisc.ernet.in

  • Venue:
  • Distributed and Parallel Databases - Security of data and transaction processing
  • Year:
  • 2000

Quantified Score

Hi-index 0.00

Visualization

Abstract

Many real-time database applications arise in electronic financialservices, safety-critical installations and military systems whereenforcing security is crucial to the success of the enterprise. Forreal-time database systems supporting applications with firm deadlines,we investigate here the performance implications, in terms of killedtransactions, of guaranteeing multilevel secrecy. In particular, wefocus on the concurrency control (CC) aspects of this issue.Our main contributions are the following: First, we identify whichamong the previously proposed real-time CC protocols are capable ofproviding covert-channel-free security. Second, using a detailedsimulation model, we profile the real-time performance of arepresentative set of these secure CC protocols for a variety ofsecurity-classified workloads and system configurations. Ourexperiments show that a prioritized optimistic CC protocol, OPT-WAIT,provides the best overall performance. Third, we propose and evaluatea novel “dual-CC” approach that allows the real-time database systemto simultaneously use different CC mechanisms for guaranteeing securityand for improving real-time performance. By appropriately choosingthese different mechanisms, concurrency control protocols that provideeven better performance than OPT-WAIT are designed. Finally, wepropose and evaluate GUARD, an adaptive admission-control policydesigned to provide fairness with respect to the distribution of killedtransactions across security levels. Our experiments show that GUARDefficiently provides close to ideal fairness for real-time applicationsthat can tolerate covert channel bandwidths of upto one bit persecond.