Non-minimum cost forwarding for packet networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network routing technologies face challenges in efficiently utilizing non-minimum cost data paths without causing forwarding loops, especially in high-speed computing networks that require balanced traffic load and congestion management.
Innovation Solution
Implementing non-minimum cost based forwarding rules that allow packets to be routed along non-minimum cost paths for multipath data transfer, ensuring that such paths are used only once before switching to a shortest path, and utilizing connectionless routing with source-aware switches to manage traffic flow across multiple hops.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If non-minimum cost paths are used for multipath data transfer, then bandwidth utilization is improved, but forwarding loops may occur
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing both minimum cost and non-minimum cost paths in forwarding tables before actual data transmission occurs. This allows switches to quickly determine appropriate paths without real-time computation, enabling efficient multipath routing while maintaining loop-free operation through pre-established routing rules.
Solution Approach 2:
The patent implements dynamics by enabling switches to dynamically select between minimum cost and non-minimum cost paths based on real-time network conditions, traffic patterns, and load balancing requirements. This dynamic path selection allows the network to adaptively utilize available bandwidth while preventing forwarding loops through context-aware routing decisions.
2Ease of operation
If traffic is routed along minimum cost paths only, then forwarding simplicity is maintained, but network congestion occurs
Solution Approach 1:
The patent applies segmentation by dividing the routing decision into two distinct segments: minimum cost path selection for normal traffic and non-minimum cost path selection for load balancing scenarios. By maintaining separate forwarding table entries for each path type, the system preserves simple minimum cost forwarding while enabling sophisticated congestion management through selective use of alternative paths.
Solution Approach 2:
The patent implements parameter changes by modifying the cost metric used for path selection based on network conditions. Switches can change from using static minimum cost metrics to dynamic metrics that incorporate load balancing considerations, allowing the network to transition between simple forwarding mode and congestion-aware mode as needed.
3Adaptability or versatility
If non-minimum cost forwarding is enabled, then path diversity increases, but routing complexity increases
Solution Approach 1:
The patent reduces routing complexity through preliminary action by pre-computing all viable paths (both minimum and non-minimum cost) and storing them in forwarding tables before runtime. This eliminates the need for complex real-time path computation and selection algorithms in switches, allowing them to simply lookup and forward packets based on pre-determined paths, thus maintaining low device complexity while achieving high path diversity.
4Productivity
If multiple paths are used for traffic flow, then bandwidth utilization improves, but loop detection difficulty increases
Solution Approach 1:
The patent converts the potential harm of using multiple paths (which could create loops) into a benefit by establishing predetermined rules that define exactly when and how non-minimum cost paths should be used. These rules, embedded in the forwarding logic, ensure that path selection never creates loops while still allowing extensive use of alternative paths for load balancing, thus eliminating the detection problem entirely through preventive design.
Data Source
AI summary
In some examples, a network node receives a packet from an adjacent node in a packet-switched network. The receiving node can forward the packet to a destination node via a minimum cost forwarding node adjacent to the network node or to a non-minimum cost forwarding node adjacent to the network node based on routing criteria for the packet-switched network. The routing criteria can include whether the adjacent node that sent the packet to the receiving node is a non-minimum cost node between a source node and the destination node for the packet.


