Resource aggregation for fault tolerance in integrated services networks

Publisher Website
Google Scholar
Abstract
For several real-time applications it is critical that the failure of a network component does not lead to unexpected termination or long disruption of service. In this paper, we propose a scheme called RAFT (Resource Aggregation for Fault Tolerance) that guarantees recovery in a timely and resource-efficient manner. RAFT is presented in the framework of the Reliable Back-bone (RBone), a virtual network layered on top of an integrated services network. Applications can request fault tolerance against RBone link and node failures. The basic idea of RAFT is to setup every fault tolerant flow along a secondary path that serves as a backup in case the primary path fails. The secondary path resource reservations are aggregated whenever possible to reduce the overhead of providing fault tolerance. We show that the RSVP resource reservation protocol can support RAFT with simple extensions.

This publication has 10 references indexed in Scilit: