An algorithmic model is presented to limit the congestion in end-to-end TCP networks, an evaluation of this type of current networks of this type is carried out and the present shortcomings are determined. An investigation is made about the metrics that are most important and determinant at the time of end-to-end TCP transmissions such as segment size, buffer capacity, ACK times and concurrent services and how they should be monitored and configured in order to obtain the best result under certain previously identified conditions of the network. The model is designed and presented as a series of steps to follow according to the actual factors that an end-to-end network presents as well as the implementation of its final design, using the same methodology during the testing and the simulation carried out with the ns2 software. Some simulations of the encountered scenarios are presented in comparison with the results of an actual end-to-end network, in order to determine the main obtained result. A series of recommendations are made and the conclusions drawn by the research are listed as well as some considerations on future work.