Abstract:

MPLS using RSVP-TE (MPLS/TE) offers three forms of restoration. These are in order of fastest restoration: 1) restoration initiated at the point of failure using local-protect or fast-reroute (FRR), 2) restoration initiated at the ingress using presignaled backup LSPs (these backups referred to here as "standby" LSPs), and 3) restoration at the ingress requiring new LSPs to be signaled after the failure is detected. The slower techniques require less LSPs to be signaled and are perceived as more scalable, requiring less bandwidth and easier to manage, though this may not be an accurate perception for reasons explained in the presentation. Regardless of the method of restoration, the duration of resource outage will generally exceed the restoration time by many orders of magnitude, requiring that primary paths be rerouted. Rerouting dynamics have been discussed in prior MPLS Conference presentations. With greater emphasis being placed on fast restoration and fast convergence, initial traffic layout efficiency suffers, and computational scaling can become an issue.

This presentation describes many aspects of restoration and convergence dynamics, including restoration and convergence times, traffic layout efficiencies, and scaling. Some implementation and network architecture and configuration techniques are described which improve performance. Simulation and test results are provided as a means to quantify the problems and the impact of various approaches.

Curtis Villamizar has been involved in Internet operations, and protocol design and implementation since working for ANS in 1992-1997 in support of the NSF funded T3-NSFNET project and later ANSNET. In 1997-1999 Curtis was part of the UUNET Network Architecture Group. In 1999 Curtis joined Avici where he is presently Principal Design Engineer and responsible for Avici's MPLS/TE implementation.

Rajesh Khanna is currently a Principal Test Engineer with Avici Systems in Billerica, MA where he has been leading a team of MPLS test engineers for over 3 years. Rajesh is responsible for evaluating new MPLS technologies for performance and network deployment issues. Prior to joining Avici, Rajesh was a Systems Engineer with HCL Comnet in India. Rajesh holds a BS in Electronics and Telecommunications from Univeristy of Pune.