| MPLS has been envisioned as an ideal platform upon which various service quality features can be developed, including Quality of Service (QoS), Traffic Engineering (TE), and Service Restoration (SR). To this end, a whole suite of protocols are being developed to enable individual features in an MPLS network, e.g., DiffServ for QoS, constrained-based routing for TE, and path protection for SR. This, however, raises serious concerns about the scalability and complexity involved in managing and operating a complex network.; In this thesis, we develop an integrated architecture to provide rich, scalable QoS, TE, and SR features simultaneously in an MPLS network. This architecture combines a resource provisioning mechanism and a distributed multipath forwarding paradigm at the MPLS domain edge to enable optimal, QoS-aware aware TE and SR through dynamic load balancing and rate adaptation. The only needed information from the network core is the number of congested links in the forwarding path.; A baseline realization of this architecture presented in this thesis enables the Best Effort and Expedited Forwarding service class and allows fully automatic multipath setup and update without human intervention. The core nodes are QoS agnostic and use a single FIFO queue at each output port. Moreover, it allows 100% optimal SR for the Expedited Forwarding class with respect to single link failures. The simulation results demonstrate the viability of the proposed architecture in providing these features without having to implement any new protocols in an MPLS network. |
