IMPLEMENTATION AND PERFORMANCE EVALUATION" ON OF MPLS ROUTER ON INTEL • I 1200 NETWORK PROCESSOR

Abstract

Network processors are programmable processors that are optimized to implement packet processing functions in a network. This work illustrates the process of implementing a Multi Protocol Label Switching (MPLS) router on Intel IXP1200 network processor and examines its performance. MPLS is an Internet Engineering Task Force (IETF) standard, which introduces a new forwarding paradigm for IP networks. In IP based routing, IP router at each hop consults the routing table for forwarding every packet. MPLS Router improves performance by forwarding packets based on a short fixed length label without address resolution at layer 3. An MPLS router can work as Ingress Label Edge Router (LER), Label Switched Router (LSR) or Egress Label Edge Router in an MPLS domain. Ingress LER is placed at the edge of an MPLS domain. It receives unlabeled IP packets, labels them and sends them to an MPLS next hop. LSR is placed in the middle of an MPLS domain. It receives labeled packets, swaps the labels and sends the packets to an MPLS next hop. Egress. LER is placed at the edge of an MPLS domain. It receives labeled packets, strips off the labels and sends the packets to an IP next hop. In this dissertation, performance of MPLS router as LERs and LSR has been compared with traditional IPv4 router. LSR and Egress LER have shown better throughput and latency. However, the performance of Ingress LER declines due to overheads of packet classification. Thus, for a large MPLS domain with many LSRs in it, the overall performance will improve