Maximally Redundant Trees for IP Fast Reroute
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional link state routing protocols, such as OSPF and IS-IS, experience significant delays in adapting forwarding tables to network topology changes due to link or router failures, leading to prolonged convergence times that negatively impact real-time applications like VoIP and multimedia services, which demand rapid failure detection and recovery.
Innovation Solution
Implementing maximally redundant trees (MRTs) for fast reroute of IP and LDP traffic, where network devices compute pairs of MRTs and pre-install alternate next-hops to ensure continuous traffic flow even during primary next-hop failures, without requiring additional signaling or external computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional link state protocols are used for routing, then routing information is exchanged and accumulated to construct network topology, but convergence time becomes excessively long (500 milliseconds to several seconds) when link failures occur
Solution Approach 1:
The patent pre-computes maximally redundant trees (MRTs) and installs alternate next-hops in advance before failures occur. When a link failure happens, routers immediately switch to pre-computed alternate paths without waiting for traditional convergence, achieving failure recovery in around 50 milliseconds
Solution Approach 2:
The patent segments the network topology into two maximally redundant trees (blue MRT and red MRT) that are computationally derived from the link state database. This segmentation allows independent failure recovery paths to be pre-established, enabling fast reroute without full topology recomputation
2Loss of time
If alternate next-hops are pre-installed to reduce failure recovery time, then convergence speed improves, but device complexity and state requirements on routers increase
Solution Approach 1:
The maximally redundant trees serve multiple functions: they provide primary forwarding paths, alternate recovery paths, and enable fast reroute for both unicast and multicast traffic. This multi-functionality reduces the need for separate dedicated backup structures, minimizing additional state requirements
Solution Approach 2:
The patent changes the structural parameters of the forwarding database by organizing next-hops according to MRT topology rather than traditional shortest-path trees. This reorganization allows efficient storage and retrieval of alternate paths without proportionally increasing state requirements
3Reliability
If maximally redundant trees are computed for all network devices, then fast reroute capability is achieved, but computational overhead and signaling requirements increase
Solution Approach 1:
Routers autonomously compute their own MRT next-hops using locally stored link state database information. Each router independently determines alternate paths without requiring external computation services or complex inter-router signaling, reducing overall system overhead
Solution Approach 2:
The patent extracts the essential MRT computation logic from complex external systems and implements it as simplified local algorithms that use existing link state data. This extraction reduces computational overhead by leveraging already-available topology information rather than requiring new computation frameworks
Data Source
AI summary
Network devices provide Internet Protocol (IP) and Label Distribution Protocol (LDP) fast reroute for unicast and multicast traffic. The approach described herein for fast reroute for IP and LDP uses maximally redundant trees (MRTs). MRTs are a pair of trees where the path from any node X to the root R along the first tree and the path from the same node X to the root along the second tree share the minimum number of nodes and the minimum number of links. A network device, such as a router, computes a pair of MRTs for each destination and installs one or more MRT alternate next-hops in its forwarding plane for use in forwarding network traffic to a destination in the event a failure occurs that renders a primary next-hop unusable for reaching the destination.