Distinguish Between Mpls-based Traffic Engineering Services

With the application of computer in our lifetime, more and more people are now connecting Internet to get to be serviced. Different people want to use Internet because of their different requirement of Internet to provide different Quality of Service (QoS). Several Techniques of QoS such as IntServ, DiffServ, MPLS and Traffic Engineering are discussed in this paper at first.IETF has proposed two models of QoS: IntServ and Diffserv. IntServ uses RSVP to reserve resource along the transfer path before traffic was send. IntServ can provide absolute end to end QoS. DiffServ defines a few different service classes. Every class corresponds to different transfer mode. This two service model has its each advantage and disadvantage. Due to the flexible extensions of DiffServ, more applications are prefer using DiffServ.At present data was transferred in the computer network along the path calculated by SPF. But the SPF algorithm can cause unbalanced traffic load among the transfer paths, some links can be overload, while others can be under load. The solution of this problem is Traffic Engineering. Traffic Engineering has two aspects. On one hand, Traffic Engineering is using Traffic Engineering algorithm to find path to avoid link overload, on the other hand, there need a frame work to unify planning and distribution of network resource.As a switch technique, MPLS has many advantages, especially in Traffic Engineering. By using explicit LSP MPLS can deploy Traffic Engineering very easily. At the same time by mapping DiffServ DSCP to MPLS label or to MPLS LSP, MPLS is applicable for DiffSev. MPLS, DiffServ and Traffic Engineering can combine together to realize QoS.Traffic Engineering algorithms are also discussed in this paper, MIRA and PBR are evaluated. MIRA defines some "Critical Link" and assign weight to these links in path calculation so as to avoid some traffic transfer through certain links. But MIRA requires heavily calculation of router. Based on study of Traffic Engineering algorithms this paper proposed a new algorithm: Parallel Path Max Traffic Flow Algorithm. This algorithm isto find max traffic flow between ingress and egress according For-Fulkerson's algorithm of Min-Cut and Max-Flow off line. There can be several combined path of Max-Flow between a certain pair of ingress and egress. We defined a interfere value of a Max-Flow Path. Comparing the interfere of the different combined Max-Flow Path, we can choose the combined Max-Flow Path with minimum interfere, and allocate this resource to different QoS level requests.Another key point of this paper is Traffic Engineering Frame Work. Current prevalent frame works such as : TEQUILA, MESCAL> TEAM were studied. A revised Traffic Engineering Frame Work was proposed: Resource Requirement Planning Traffic Engineering Frame Work. Overall network resource and traffic requests were taken into account. Traffic request was scheduled according the current network resource planning to avoid traffic congest preventively.Finally, Parallel Path Max Traffic Flow Algorithm's Simulation was carried out using NS 2.27. Three Scenarios were considered: No Traffic Engineering and No DiffServ, No Traffic Engineering and with DiffServ, with Traffic Engineering and with DiffServ. Simulation generates data of QoS performance: loss and throughout of these three Scenarios. Analysis of the Simulation result showed that QoS performance was significantly improved.