Weighted Cost Multi-Path Flow Distribution for Network Congestion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Network congestion and increased latency due to packet collisions caused by multiple data packets requesting the same destination, leading to inefficient bandwidth utilization in multi-path networks.
Innovation Solution
Implementing Weighted Cost Multi-Path (WCMP) routing with iterative scaling of weights in forwarding tables to reduce the number of entries and minimize throughput degradation, using a processor to adjust weights proportionally until a predetermined event or threshold is reached.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If WCMP groups are implemented with multiple forwarding table entries for load balancing, then network throughput is improved, but the number of forwarding table entries increases
Solution Approach 1:
The patent applies parameter changes by iteratively scaling the weights assigned to ports in WCMP groups. The processor adjusts weight parameters (W0, Wi) proportionally to reduce the number of forwarding table entries while maintaining load balancing effectiveness. This is achieved by decrementing weights using the equation Wi(G′)−=floor([W0(G)−W0(G′)]*Wi(G)/W0(G)) until a predetermined event occurs or throughput degradation threshold is reached.
2Reliability
If more forwarding table entries are used for precise load balancing, then packet collision avoidance is improved, but hardware resource consumption increases
Solution Approach 1:
The patent reduces hardware resource consumption by dynamically adjusting weight parameters in the forwarding table. The processor iteratively scales weights to find an optimal configuration that maintains sufficient load balancing to avoid packet collisions while using fewer forwarding table entries. This parameter optimization reduces the replication proportion of WCMP group members in the forwarding table.
Solution Approach 2:
The patent applies partial action by not requiring full-weight replication of all WCMP group members. Instead, it selectively scales down weights of less critical ports while maintaining adequate load balancing for high-traffic paths. This partial optimization reduces the total number of forwarding table entries while preserving essential collision avoidance capabilities.
3Device complexity
If forwarding table entries are reduced, then hardware resource consumption is decreased, but throughput degradation may occur
Solution Approach 1:
The patent implements feedback control by continuously monitoring network throughput and using this information to guide the weight scaling process. The processor adjusts weights iteratively and stops when a predetermined event occurs or when the throughput degradation reaches a predefined threshold. This feedback mechanism ensures that hardware resources are optimized while maintaining acceptable network performance levels.
Data Source
AI summary
The present technology considers network devices that include forwarding tables having a number of next-hop entries (e.g., egress ports) where it is possible that the egress port utilization can be load balanced using WCMP groups. To implement a WCMP group, its members are assigned weights representing an amount of data flow to distribute over a plurality of links for a given destination. The members may be replicated in a forwarding table proportionally to their assigned weights. This disclosure provides systems and methods to reduce the number of forwarding table entries required when implementing a WCMP group.