Presently, the overwhelming majority of scientific and industrial experts hold that the assessment of data availability plays a crucial role in the security and safety assurance for any of software and hardware complexes. Still, for a number of industrial facilities, especially for control systems, the problem of availability assessment has been not yet properly solved. In this paper, an original approach to the availability assessment is presented: the maximum delay of data transmission is elected as a measure of availability, and a theory of deterministic queuing systems Network calculus is used for its calculation. The paper uncovers the advantages of Network calculus and describes how to apply it. Using the approach, an investigation of IntServ model availability under malicious actions by digital means, or cyberattacks, is carried out. The dependence of delays on arrival flow and server parameters is discovered. As a part of the proposed approach, a technique for computing linear bounds from empirical data for an arrival process and service curve is discussed.