Thursday, May 13, 2010 9:00 am - 9:45 am
When deploying multicast services over a Point-to-multipoint (P2MP) MPLS-TP or MPLS-TE network it is important to identify a suitable route between the source and a set of destinations. This planning process for the service path can be subject to a set of constraints including: bandwidth, latency, QoS, protection, and policy. Performing multi-constraint path computation for multicast services within a single domain can pose significant computational and planning problems.
When computing paths for inter-domain multicast services, a number of additional challenges are raised including how to determine the optimal sequence of domains traversed by the path. The Path Computation Element (PCE) enables administrators of separate domains to participate in the selection of optimum end-to-end paths without compromising the confidentiality of their topologies or the state of their network resources. Paths can be routed between domains that are multiply interconnected, and policy can be applied to determine issues like the placement of branching points to make trade-offs between least cost trees and reduction of domain border transits. Thus the computation of P2MP paths may be made subject to a variety of real commercial as well as service-oriented constraints and considerations.
In this presentation we demonstrate how PCE technology can be combined with network planning techniques and systems to achieve optimal delivery of high-capacity multicast services over a multi-domain packet network. A PCE-enabled solution facilitates rapid and scalable selection of optimal paths across an inter-domain environment, and can take advantage of both typical traffic engineering constraints and more commercially relevant constraints such as policy, SLAs, security, peering preferences, and dollar costs. We will also share test results, which compare a PCE-enabled architecture versus more traditional techniques, for deploying high capacity inter-domain multicast services.