In this work, real traffic of a service provider IP/MPLS network is classified per application using a traffic characterization procedure based on self-similarity. The results of such classification show that the real traffic confirms a high degree of burstness due to high values of the Hurst parameter. These results are used for the proposal of a method of network capacity planning. The major difficulty is the knowledge of the traffic inside the backbone, and this needs to be a self-learning process mainly because every day new service applications are inserting new type if traffic. In this work an ILEC(incumbent local exchange carrier) service provider IP traffic will be classified per application after that applying a self similarity characterization some analysis will be done.