Telecommunication network topologies are dynamically changing nowadays. The core telecommunication networks are based on fiber optics infrastructure. WDM technology is used to transmit multiple flows of traffic over a single fiber. Virtual network topology (VNT) is used to route IP traffic over WDM networks. VNT control system must be adaptive to traffic changes and dynamically reconfigure VNT. This can be achieved by using attractor selection, which models biological system behavior. Biological systems are extremely adaptive to unknown changes in environmental conditions. This adaptability could be used as a model for VNT control system. Such VNT control method is based on a system of stochastic differential equations, which use load on links, retrieved by SNMP, and stochastic fluctuations – Gaussian noise – as input parameters to dynamically reconfigure VNT and optimize its performance. Network simulations show that this VNT control method is very quick and adaptable to traffic changes.