NIC Port Virtualization for Fast Network Failover
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large data centers experience low fault tolerance and computationally costly restarts due to NIC transceiver failures, particularly in AI workloads, leading to downtime and productivity loss.
Innovation Solution
Implementing a software stack solution for dataplane control that rapidly switches network connections from a failing port to a reserve port using a NIC with onboard processing capabilities, updating network forwarding rules to ensure seamless continuity without full reset.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If virtualized services are migrated to different underlying hardware when port failures are detected, then fault tolerance is improved, but computational overhead and restart costs increase significantly for AI workloads
Solution Approach 1:
The system pre-establishes backup network paths and standby hardware resources before failures occur. When a port failure is detected, the virtualized service can immediately switch to the pre-prepared backup path without requiring computational restart or migration, thus improving fault tolerance while avoiding high computational overhead.
Solution Approach 2:
The patent introduces a network controller as an intermediary that manages failover logic and path selection. This intermediary handles the complexity of failure recovery, allowing the AI workload to continue running while the controller transparently redirects traffic through alternative paths, eliminating the need for computationally expensive service restarts.
2Reliability
If traditional network reconfiguration is performed after port failure, then connectivity is restored, but downtime and productivity loss occur due to the reconfiguration process
Solution Approach 1:
Backup network paths are pre-configured and standby hardware resources are prepared in advance before any failure occurs. When a port failure is detected, the system can immediately activate the pre-prepared backup path, restoring connectivity instantly without requiring time-consuming reconfiguration processes.
Solution Approach 2:
The system implements dynamic path selection and adaptive failover mechanisms that can rapidly respond to failures. The network controller continuously monitors port status and can dynamically switch between primary and backup paths in real-time, minimizing downtime by making rapid reconfiguration decisions rather than performing traditional sequential reconfiguration.
3Stability of the object's composition
If full network reset is performed to restore connectivity after failure, then network stability is achieved, but operational continuity is interrupted causing productivity loss
Solution Approach 1:
The patent extracts the failover logic from the AI workload itself and places it in a dedicated network controller. This allows the network to perform localized failover at the controller level without interrupting the AI computation, maintaining both network stability and operational continuity simultaneously by isolating the recovery process from the workload.
Solution Approach 2:
The network controller acts as an intermediary that maintains network stability through intelligent path selection without requiring full network resets. It can selectively switch individual failed paths while keeping the rest of the network and AI workloads running, thus preserving operational continuity while achieving the necessary stability through targeted rather than comprehensive resets.
Data Source
AI summary
A network resiliency controller may monitor a port status for a network interface controller. A software defined datapath may be used to virtualize different ports of the network interface controller and direct traffic to a given port. If it is determined that a port failure is imminent or has occurred, one or more selectors may modify a port associated with the network interface controller. The network resiliency controller may identify the port switching, determine a new connection has been established with a new port, and then modify one or more traffic rules for routing traffic along the new port.