The work established that an essential solution for proactively ensuring fault tolerance of networks is the support of load balancing both at the transport network level and access level using FHRP. However, FHRP load balancing is based on manual settings, which impose high requirements on the network administrator’s professional training and experience level. Therefore, the task of improving mathematical models and methods that make up the algorithmic basis of fault-tolerant routing protocols is urgent. At the same time, a mandatory requirement for these models and methods is to consider the border routers’ reliability through which the load incoming from access networks is balanced. The work describes four mathematical solutions to the problem of proactive fault-tolerant routing. To ensure a high level of Quality of Service, all analyzed solutions support the requirements of the Traffic Engineering concept, and two take into account the reliability of border routers (RATE and ResMetrTE). On the network topology chosen for the study, the problem of proactive fault-tolerant routing was solved using the solutions described in work. The results of the calculations confirmed the sensitivity of the RATE and ResMetrTE routing solutions to the reliability of border routers. Within the considered example, it was established that taking into account the level of border routers’ reliability when organizing load balancing between them using RATE or ResMetrTE solutions led to an increase in the upper bound of the network link utilization – from 15% to 27% on average. The work demonstrates that the implementation of analyzed load balancing solutions can be ensured using the GLBP protocol using the weighted balancing mode when the weight of each border router is determined not empirically but based on the results of calculations within the RATE or ResMetrTE solutions.

A practical approach to load balancing in a telecommunication network (TCN) is implementing Traffic Engineering (TE) technology principles to reduce link utilization and improve QoS level. In order to adapt TE solutions with network security requirements, this paper proposes a mathematical model for secure routing, which belongs to the class of flow-based optimization solutions. The model is based on the conditions of multi-flow routing implementation, flow conservation, and TCN link overload prevention. Due to this, the problem of secure routing is formulated in an optimization form. The model’s novelty is the modified conditions of load balancing in TCN. Along with the indicators of link capacity with the help of weighting coefficients, the network security (NS) indicators of TCN elements are also taken into account. The network security (NS) indicators in the TCN modeling process include information security risks of routers and communication links, losses from breach of confidentiality and integrity of information, probability of existing vulnerabilities exploitation, etc. The study confirmed the effectiveness of the proposed solution. On the test TCN topology, it is demonstrated that the use of a secure routing model allows to calculate the routes and provide such an order of load balancing, which compromises meeting the requirements of both QoS and NS. In the routing process, information security risk reduction in packet transmission by about 11.3% was accompanied by an increase (on average by 26%) in the upper bound of the network link utilization