Routing is an indispensable network function that connects a call/session from origin to destination, and is at the heart of the architecture, design, and operation of any network. The introduction of dynamic routing into telecommunications networks has resulted in dramatically improved network robustness, particularly in responding to abnormal traffic or failure conditions, while simultaneously reducing network costs. Voice network dynamic routing has been in operation for 20+ years and is deployed in many large-scale operational networks worldwide. We describe various dynamic routing implementations, including time-dependent routing (TDR), state-dependent routing (SDR), and event-dependent routing (EDR). We trace dynamic network evolution from hierarchical routing networks to real-time dynamic networks and beyond, and illustrate dynamic network performance under various conditions of network stress. Significant lessons have been learned in terms of dynamic routing principles to ensure efficient performance for large-scale networks, which include connection admission control for bandwidth allocation, bandwidth reservation for stable performance, EDR path selection for reduced routing overhead, class-of-service mechanisms for multi-service bandwidth allocation, and dynamic transport routing for fast, automated network restoration. We show that these lessons and principles can be applied to converged voice/data networks employing multiprotocol label switching (MPLS) and generalized MPLS (GMPLS), and illustrate traffic engineering and QoS optimization (TQO) methods applicable to packet flow routing, MPLS/GMPLS label switched path (LSP) routing, and dynamic transport routing functions in such networks. We present analysis studies that illustrate the tradeoffs between various TQO approaches, and provide detailed case studies of the optimization of MPLS/GMPLS-based integrated voice/data dynamic routing networks. The benefits of such MPLS/GMPLS architecture evolution are quantified in terms of lower operations and capital costs, improved network performance, and simplified/automated network management.
Group for Research in Decision Analysis