The article describes an approach to the numerical analysis of the stability of distributed information systems under a load that varies over time allowing for a short-term load exceeding critical values. Information systems are considered as imperfect queuing systems in which the intensity of the message flow may depend on the size of the queues of applications. Based on this approach, criteria for the stable and unstable behavior of the system under load are formulated, where the stable behavior of the system is understood as the ability of the system to return to its normal mode of operation independently when the external load drops below the critical value, and unstable one when there is the transition to an emergency state. As examples, the most common types of unstable systems are considered, as well as an example of a distributed system that implements a complex queuing system, which, depending on the characteristics of the service, can be either stable or unstable. The analysis of such a system is carried out, which is confirmed by the results of numerical simulation. Based on the analysis results, zones of stability and instability are determined. Features of the implementation of distributed real-time information systems are considered, practical recommendations for the implementation of sustainable information systems are given
Read full abstract