Flexible Ethernet Logical Lane Aggregation for Link Resilience
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-speed Ethernet datacenter interconnect links face complete failure when any individual lane fails, as they require all lanes to be operational, leading to inefficiencies and increased operational costs due to non-field-serviceable optics.
Innovation Solution
The implementation of Flexible Ethernet Logical Lane Aggregation (FELLA) allows for selective removal or addition of lanes from service, redistributing data across remaining operational lanes without altering the media access controller or replacing optical modules, enabling continued operation at reduced bandwidth and improving fault tolerance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all lanes must be operational for the interconnect link to function, then link reliability is improved, but system resilience deteriorates because a single lane failure causes complete link failure
Solution Approach 1:
The interconnect link is segmented into multiple independent lanes that can operate independently. Each lane can be individually enabled or disabled without affecting the operation of other lanes, allowing the system to maintain functionality even when some lanes fail.
Solution Approach 2:
The system dynamically adjusts the number of active lanes based on operational conditions. Lanes can be selectively enabled or disabled during operation to adapt to failures or changing bandwidth requirements, transforming a static all-or-nothing link into a dynamic flexible system.
2Use of energy by moving object
If lanes are removed from service to save power, then energy efficiency is improved, but device complexity increases due to lane management requirements
Solution Approach 1:
The system provides dynamic power management by allowing lanes to be selectively enabled or disabled based on actual bandwidth requirements. This enables the interconnect link to consume only the necessary amount of power for the current workload, improving energy efficiency.
Solution Approach 2:
The system includes automatic detection and lane management capabilities that can identify failed lanes and redistribute traffic without requiring manual intervention or complex external management, reducing the operational complexity burden.
3Ease of manufacture
If non-field-serviceable optics are used, then manufacturing cost is reduced, but ease of repair deteriorates because failed lanes cannot be individually replaced
Solution Approach 1:
The system treats each lane as a separatable logical unit that can be individually managed, enabled, or disabled. This logical segmentation compensates for the physical non-separability of integrated optics, allowing failed lanes to be isolated and their traffic redistributed without replacing entire optical modules.
Solution Approach 2:
When a lane fails, the system automatically discards traffic on that failed lane and recovers by redistributing the traffic across remaining functional lanes, maintaining system operation without requiring physical replacement of failed components.
4Measurement precision
If lane failures cause complete link failure, then measurement precision of link status is improved, but loss of time increases due to full re-initialization requirements
Solution Approach 1:
The link status monitoring is segmented at the lane level, allowing individual lanes to be monitored and identified as failed independently. This granular detection enables selective lane disabling rather than complete link shutdown, dramatically reducing recovery time.
Solution Approach 2:
The system performs preliminary detection and isolation of failed lanes, automatically disabling only the affected lanes before traffic redistribution is required. This preliminary action prevents complete link failure and eliminates the need for full re-initialization sequences.
Data Source
AI summary
A system or method for logical lane aggregation provides data across a first set of interconnect lanes and determines that at least one interconnect lane is unavailable. The system or method redistributes data to a second set of interconnect lanes, the second set not including the at least one interconnect lane, in response to the at least one interconnect lane being unavailable. The system or method can be used to provide flexible Ethernet logical lane aggregation (FELLA).