Interconnect Module Smart NIC Offload
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In software-defined infrastructure, the repeated routing of network packets through two layers (ICM and hypervisor) leads to computational overhead, causing performance degradation for virtual machines during high network traffic, as the hypervisor shares finite CPU resources between network handling and VM computations.
Innovation Solution
Implementing Smart NICs (SNICs) that offload network switching and routing functions from the hypervisor to dedicated processing units within the NIC, reducing the load on the ICM and hypervisor by preprocessing network traffic and providing direct switching/routing capabilities, thus enhancing CPU availability for VM workloads.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If network traffic is routed through two layers (ICM and hypervisor), then network connectivity is achieved, but computational overhead increases and performance degrades
Solution Approach 1:
The patent extracts the network switching and routing functions from the hypervisor software layer and relocates them to dedicated hardware components (Smart NICs and ICM). This extraction eliminates the computational overhead of software-based network handling, allowing the hypervisor to focus on VM management while hardware handles network traffic efficiently.
Solution Approach 2:
The patent introduces Smart NICs as intermediary devices between the VMs and the ICM. These Smart NICs include onboard switching/routing capabilities that preprocess network traffic before it reaches the ICM, acting as a mediator that reduces the processing burden on both the hypervisor and the core switching infrastructure.
2Adaptability or versatility
If hypervisor handles network switching and routing, then software-defined networking is maintained, but CPU resources are consumed for network processing
Solution Approach 1:
The patent segments network processing functions across multiple layers: L2 switching at the ICM, L3 routing at the Smart NICs, and VM management at the hypervisor. This segmentation allows each component to handle specific tasks efficiently, with hardware handling data plane operations and software handling control plane functions.
Solution Approach 2:
The Smart NICs are equipped with onboard processing units that autonomously perform network switching and routing operations without requiring hypervisor intervention. This self-service capability allows the network hardware to handle its own traffic processing, freeing CPU resources for other tasks while maintaining software-defined networking through programmable interfaces.
3Ease of operation
If basic NIC is used for converged network traffic, then network merging is achieved, but switching and routing participation is limited
Solution Approach 1:
The patent merges multiple network functions into the Smart NIC component, combining L3 routing, L2 switching, and converged network traffic handling in a single hardware device. This integration allows the Smart NIC to participate actively in network switching and routing while maintaining ease of operation for traffic convergence.
Data Source
AI summary
An interconnect module (ICM) having at least two internal data paths is provided. The ICM determines if a connected network interface card (NIC) supports a division of work between the NIC and the ICM. NICs may be standard NICs, advanced NICs (ANICs), or smart NICs (SNICs). The ICM may perform a different amount of processing for network packets received from different devices based on the division of work previously identified. Some NICs may preprocess network packets with respect to switching and routing processing to allow the ICM to bypass that functionality for those packets. Packets received from devices not providing a division of work receive full processing including switching and routing processing. Devices may be grouped to either a switching and routing group or a virtual bypass group such that data received from devices associated with the virtual bypass group may bypass selected processing typically performed by the ICM.