Service Restoration in MPLS Networks

Students: Arun Gandhi and Mythilikanth Raman
Faculty: Radim Bartoš


In this project we proposed new approaches based on the concept of domain protection, where protection of paths leading from all ingress routers to a common egress router is calculated simultaneously. The scheme allows a flexible protection path placement independent of the working path placement. Further, the scheme allows for protection of only a selected subset of traffic streams and permits operator-engineered path placement, as often required in practice. A majority of the current service restoration schemes either provide fast response time at the expense of preallocation of resources or attempt to allocate the resources after the failure is detected making it difficult or impossible to react within the desired time frame. One of the contributions in this area is the introduction of two-phase protection schemes that allow for fast protection switching while utilizing less resources. The schemes take immediate reparative actions carried out by the nodes adjacent to the failure while considering only locally-available information. This phase lasts only until full service restoration measures are calculated and signaled. The transient nature of this phase allows for lesser resource requirements, especially considering common practical scenarios where only a small portion of carried traffic requires tight guarantees. The rest of the traffic expects lower-grade service and can be temporarily subjected to degraded service as the network resources are utilized for restoration of high-priority traffic. Partial allocation of protection resources unavoidably leads to QoS degradation during protection switching. To address this issue, we developed and analyzed a mechanism based on distributed buffering in the existing queues to control the flow of traffic on the paths so that packet loss and reordering are minimized. The mechanism is general and independent of the particular protection scheme.

The long-term goal of this project is to develop technology-neutral architectures, protocols, and procedures for building and operating networks with high availability.


  • R. Bartoš and A. Gandhi, ``Dynamic issues in MPLS service restoration.'' Proc. of the Fourteenth International Conference on Parallel and Distributed Computing and Systems (PDCS), Cambridge, MA, (S. G. Akl and T. Gonzalez, eds.), pp. 618-623, November 2002. [PDF 91k]
  • R. Bartoš and M. Raman, ``A heuristic approach to service restoration in MPLS networks,'' in Proc. of the 2001 IEEE International Conference on Communications (ICC), Helsinki, Finland, June 2001. [PDF 86k]
  • R. Bartoš, M. Raman, and A. Gandhi, ``New approaches to service restoration in MPLS-based networks,'' in Proc. of the IEEE International Conference on Trends in Communications (EUROCON 2001), Bratislava, Slovakia (P. Farkas ed.), pp. 58-61, July 2001. [PostScript 141k]
  • R. Bartoš and M. Raman, ``A scheme for fast restoration in MPLS networks,'' in Proc. of the Twelfth International Conference on Parallel and Distributed Computing and Systems (PDCS), Las Vegas, NE (M. Guizani and X. Shen, eds.), pp. 488-493, November 2000. [PostScript 282k]