In-Band Lane Failover Protocol for High-Speed Serial Links
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-performance computing systems face interruptions and network congestion during lane failover in high-speed serial interconnects due to signal loss and error rate issues, leading to reduced performance and potential system failures.
Innovation Solution
An in-band real-time physical layer protocol that monitors lane status and performs seamless failover operations by keeping all lanes online, allowing continuous service even when a lane fails, with simultaneous support for transitions between different power states and data formats, minimizing downtime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all lanes are removed from service during fail over to re-initialize to reduced width, then reliability is improved by eliminating failing lanes, but productivity deteriorates due to network traffic re-routing and buffering causing congestion and performance reduction
Solution Approach 1:
The patent segments the communication link into individual lanes that can be independently managed. When a lane fails, only that specific lane is removed from service while other lanes continue to operate, rather than removing all lanes. This allows selective maintenance of functional segments while preserving overall system productivity.
Solution Approach 2:
The patent implements preliminary monitoring and detection of lane errors before complete failure occurs. By detecting error rates approaching unacceptable levels and initiating fail over procedures in advance, the system prevents catastrophic failures while minimizing disruption to network traffic flow.
2Reliability
If all lanes are removed from service during fail over for re-initialization, then reliability is improved by ensuring clean state transition, but loss of time increases due to complete link interruption and re-routing requirements
Solution Approach 1:
The patent applies segmentation by isolating the failing lane from the operational lanes. The non-failing lanes remain in service throughout the fail over process, allowing continuous data transmission on functional lanes while the failing lane undergoes re-initialization, thereby minimizing total downtime.
Solution Approach 2:
The patent maintains continuity of useful action by keeping functional lanes active during fail over operations. Network traffic continues to flow through operational lanes without interruption, eliminating the need for complete link shutdown and re-initialization that would cause service interruption.
3Reliability
If circuit complexity is increased to manage signal loss at higher bit rates, then reliability is improved by compensating for signal degradation, but device complexity increases with additional mitigation tools and active signal repetition
Solution Approach 1:
The patent segments the signal transmission into multiple independent lanes, each capable of operating at reduced rates when needed. This allows the system to manage signal loss by activating fewer lanes at higher rates rather than requiring complex signal repetition circuits for all lanes, thereby reducing overall device complexity.
Solution Approach 2:
The patent implements dynamic lane configuration where the number and rate of active lanes can be adjusted based on signal quality conditions. This dynamic adaptation allows the system to optimize performance without requiring fixed complex mitigation circuits, as the lane configuration changes in response to real-time channel conditions.
Data Source
AI summary
A system and method provide a communications link having a plurality of lanes, and an in-band, real-time physical layer protocol that keeps all lanes on-line, while failing lanes are removed, for continuous service during fail over operations. Lane status is monitored real-time at the physical layer receiver, where link error rate, per lane error performance, and other channel metrics are known. If a lane failure is established, a single round trip request/acknowledge protocol exchange with the remote port completes the fail over. If a failing lane meets an acceptable performance level, it remains on-line during the round trip exchange, resulting in uninterrupted link service. Lanes may be brought in or out of service to meet reliability, availability, and power consumption goals.