Typically the traffic through the network is heterogeneous and it flows from multiple utilities and applications.Considering todays threats in network there is yet not a single solution to solve all the issues because the traditional methods of port-based and payload-based with machine learning algorithm suffers from dynamic ports and encrypted application.Many international network equipment manufactures like cisco, juniper also working to reduce these issues in the hardware side.Here this paper presents a new approach considering the idea based on SOTC.This method adapts the current approaches with new idea based on service-oriented traffic classification(SOTC) and it can be used as an efficient alternate to existing methods to reduce the false positive and false negative traffic and to reduce computation and memory requirements.By evaluating the results on real traffic it confirm that this method is effective in improving the accuracy of traffic classification considerably,and promise to suits for a large number of applications.Finally, it is also possible to adopt a service database built offline, possibly provided by a third party and modeled after the signature database of antivirus programs,which in term reduce the work of training procedure and overfitting of parameters in case of parameteric classifier of supervised traffic classification.