Nowadays, a great variety of methods and models are available for software maintenance effort estimation. Each method or model provides desirable results for an appropriate type of software, maintenance technique, and estimated parameters. However, none of them is suitable for estimating databases updating time because of iterative nature of databases life cycle and routine technique for their updating. It is clear that Queueing theory is close to resolve this problem. Unfortunately, it does not meet the requirements of all kind of databases. In particular, it is not useful for critical infrastructure systems with high operational risk and for systems with indeterminate or “unknown” arrival jobs processes, which include nuclear power projects. The article revealed some weaknesses of Queueing theory when using it for nuclear power projects databases and proposed a novel approach based on Network calculus. An example of using the approach for a nuclear power plant instrumentation and control system database is also presented.