Uplink Failure Rebalancing in Multi-Homing Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multi-homing network configurations in information handling systems face inefficiencies in uplink failure handling, where downlinks are either fully operational or shut down based on the number of active uplinks, leading to suboptimal network utilization, as they fail to adapt when only a subset of links fail or when specifying the active number of uplinks results in unnecessary diversion of data traffic.
Innovation Solution
Implementing uplink rebalancing through agents operating on top-of-rack devices and smart network interface controllers, which monitor uplink status and adjust link priorities and data traffic hashing to optimize network pathways, allowing for load balancing based on link bandwidth and reducing unnecessary shutdowns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If downlinks are shut down when a minimum number of uplinks are not active, then network reliability is improved, but network utilization deteriorates
Solution Approach 1:
The patent implements dynamic load balancing by continuously monitoring uplink status and adjusting traffic distribution in real-time. Instead of static shutdown decisions, the system dynamically recalculates hash weights for affected TOR devices and redistributes traffic proportionally across remaining functional uplinks, maintaining both reliability and utilization
Solution Approach 2:
The system changes the parameter of traffic distribution by modifying hash weights assigned to different TOR devices based on their current uplink status. When uplinks fail, the hash weights are adjusted to reflect the new capacity distribution, allowing traffic to be optimally routed through available paths without complete shutdown
2Reliability
If downlinks are completely shut down when uplink failures occur, then traffic is protected from failed paths, but network adaptability deteriorates
Solution Approach 1:
The system dynamically adapts to changing network conditions by continuously monitoring uplink status and adjusting traffic distribution in real-time. When uplinks fail, the system recalculates optimal traffic paths and redistributes loads proportionally, allowing the network to adapt gracefully rather than shutting down completely
Solution Approach 2:
The patent implements a feedback mechanism where the system continuously monitors uplink status and uses this information to adjust traffic distribution. The monitoring component detects uplink failures and triggers recalculation of hash weights, creating a closed-loop system that adapts based on actual network conditions
3Stability of the object's composition
If traffic is completely diverted from affected TOR devices, then network stability is improved, but device complexity increases
Solution Approach 1:
Instead of completely diverting traffic from affected TOR devices, the system applies partial action by redistributing traffic proportionally based on remaining capacity. The hash weight adjustment applies a fraction of traffic redistribution rather than complete diversion, reducing the complexity of traffic management while maintaining stability
Data Source
AI summary
Embodiments herein facilitate the modification of data traffic load balancing on information handling systems affected by a networking information handling system having the status of one or more of its uplinks changed from operable to inoperable or from inoperable to operable. In one or more embodiments, an agent operating on or in conjunction with a networking information handling system (e.g., a TOR) detects a change in one its links. The agent sends a message to information handling system(s) (e.g., hosts) that are communicatively coupled to the TOR regarding the change in status. Based upon the TOR's message, a host may adjust its traffic load balancing to compensate for the status change. Embodiments, therefore, help efficiently utilize network pathways.