The high-value Internet services that have been significantly enhanced with the integration of network virtualization and Software Defined Networking (SDN) technology are increasingly attracting the attention of end-users and major computer network companies (Google, Amazon, Yahoo, Cisco, ...). In order to cope with this high demand, network resource providers (bandwidth, storage space, throughput, etc.) must implement the right models to understand and hold the users' needs while maximizing profits reaped or the number of satisfied requests into the virtual networks. This need is even more urgent that users' requests can be linked, thereby imposing to the InP some constraints concerning the mutual satisfaction of requests, which further complicates the problem. From this perspective, we show that the problem of resource allocation to users based on their requests is a knapsack problem and can therefore be solved efficiently by using the best dynamic programming solutions for the knapsack problem. Our contribution takes the dynamic resources allocation as a multiple knapsack's problem instances on variable value requests.