Experimental Demonstration of DREAMSCAPE with More than 1000 GMPLS-based Control Nodes

Abstract

With the development of large scale multi-layer and multi-domain optical networks, path computation becomes more complicated as complex policies and constraints are to be considered and dynamic routing becomes one of the key issues in GMPLS/ASON networks. A novel Dual Routing Engine Architecture in Multi-layer and multi-domain Scalable and Constraint-Aware Policy-Enabled optical networks (DREAMSCAPE), exploiting the advantages of both GMPLS control plane and Path Computation Element (PCE), is introduced to improve the performance of routing in multi-layer and multi-domain optical networks in this paper. In the architecture, the cooperation of Group Engine (GE) and Unit Engine (UE) makes it effective to optimize path computation especially under complex multi-constraints. The functional architectures of dual routing engines are described in the paper. Based on this novel architecture, a testbed with more than 1000 GMPLS-based control nodes has been built in our lab using virtual machine technology, and two path computation approaches have been implemented on the testbed. The latest results are experimentally demonstrated on the testbed, and some future research issues are presented in the conclusion.