NIC Mapping for Kubernetes Pods Requiring Physical NIC Functions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Virtual network interface cards in Kubernetes, while providing some network functions, do not have all the capabilities of physical network interface cards, leading to potential task failures and performance issues, especially in high-performance tasks like deep learning distributed training.
Innovation Solution
A method to determine whether a network interface card is virtual or physical and configure it accordingly to meet the required acceleration card resources for deep learning tasks, either mapping a virtual card to a physical one or optimizing its configuration based on resource needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a virtual network interface card is used in Kubernetes, then network virtualization and resource sharing are improved, but the complete functionality and performance of the original network interface card are lost
Solution Approach 1:
The patent introduces a mapping mechanism as an intermediary between virtual network interface cards and physical network interface cards. When a virtual NIC is allocated to a pod group, the system maps it to an available physical NIC through the mapping table, ensuring that the physical hardware's complete functionality is accessible to the virtual instance. This resolves the contradiction by maintaining virtualization benefits while preserving full hardware capability through the mapping layer.
Solution Approach 2:
The patent implements a universal configuration mechanism where the same virtual network interface card can be dynamically mapped to different physical network interface cards based on availability and requirements. The system universally handles both virtual and physical NICs through a unified allocation and mapping process, allowing the virtual NIC to inherit all necessary functions from its physical counterpart while maintaining virtualization advantages.
2Extent of automation
If SR-IOV virtualization is implemented, then network function virtualization is achieved, but high-performance network functions are not fully provided
Solution Approach 1:
The mapping mechanism serves as an intermediary that bridges the performance gap between virtualized and physical network interfaces. By establishing direct mappings between virtual NICs and physical NICs, the system enables high-performance network functions to be fully provided while maintaining automated virtualization. The mapping table allows performance-critical traffic to bypass virtualization overhead by directly accessing the physical hardware.
Solution Approach 2:
The patent changes the state of network interface cards between virtual and physical forms through dynamic mapping. When a virtual NIC needs high performance, the system changes its state by mapping it to a physical NIC, allowing the network function to operate at full power. This parameter change enables the system to adapt between virtualization mode and high-performance mode as needed.
3Adaptability or versatility
If virtual network interface cards are allocated to pod groups, then resource sharing is enabled, but acceleration card resources may be insufficient for deep learning tasks
Solution Approach 1:
The mapping mechanism acts as an intermediary that ensures sufficient acceleration card resources are allocated to pod groups executing deep learning tasks. When a virtual NIC is mapped to a physical NIC, the mapping table tracks and guarantees that the physical hardware's acceleration capabilities are adequately provisioned. This allows the system to share virtualized resources while ensuring that resource-intensive deep learning workloads receive sufficient acceleration capacity.
Data Source
AI summary
The present application relates to a network interface card configuration method and apparatus, an electronic device, and a storage medium. The method includes: associating a target pod group with a target network interface card identifier during generation of the target pod group; obtaining a first required acceleration card resource corresponding to the target pod group in response to the fact that the target network interface card corresponding to the target network interface card identifier is the target virtual network interface card; and according to the first required acceleration card resource, configuring the target virtual network interface card for the target pod group, or mapping the target virtual network interface card to the target physical network interface card corresponding to the target virtual network interface card, and configuring the target physical network interface card corresponding to the target virtual network interface card for the target pod group.