Analysis of formal models for ensuring data integrity and their applicability to databases

Abstract

Information systems in general and databases in particular are vulnerable to accidental or malicious attacks aimed at compromising data integrity. Security is easier if you have a clear model that is the formal expression of security policy. The paper explores known security models related to data integrity, their applicability and significance for databases. The analysis of formal models for ensuring data integrity revealed that each of them, having certain advantages and disadvantages, has the right to use. The decisive factor in making a decision is an assessment of a specific situation, which will make it possible to make the right choice, including their complex application. In this regard, the paper notes that the Clark-Wilson model, the undoubted advantages of which are its simplicity and ease of joint use with other security models, is advisable to use as a set of practical recommendations for building an integrity assurance system in information systems. While stating the fact that traditional DBMSs support many of the mechanisms of the Clark-Wilson model, the article points out that implementations based on standard SQL require some compromise solutions. Analyzing the Biba model, the paper concludes about its relative simplicity and the use of a well-studied mathematical apparatus. It is noted that in practice, for the creation of secure information systems, as systems that ensure the confidentiality and data integrity, it is important to unite the Bell-LaPadula and Biba models. Moreover, this union should be on the basis of one common lattice, but with two security labels (confidentiality and integrity) with the opposite character of their definition. This is exactly the variant of combining the Bell-LaPadula and Biba models that is recommended for use in modern information systems and DBMSs, where a mandatory security policy is implemented.