A Hierarchical LSP Management Architecture for MPLS Traffic Engineering

In this paper, we propose a scalable Label Switched Path (LSP) management architecture for Multiprotocol Label Switching (MPLS) traffic engineering using a hierarchical network model. MPLS introduces the concept of a label hierarchy to support scalability by lessening the complexity of transit routers for the computation of complete routing tables. However, this hierarchical network model creates a critical disadvantage in providing a globally-optimal route because of topology abstraction or aggregation. This paper proposes a hierarchical routing scheme that can provide a globally-optimal route in hierarchical MPLS networks using propagation from the Condensed Subordinate Route Information (CSRI). CSRI is summarized route information among border nodes in the lower layer network and is reflected in the process of the LSP path computation in the higher layer network. We also propose the algorithms for generation of CSRI, for reflection CSRI to higher layer network topology, and for computation of an optimal LSP in a higher layer.