Smart NIC Leader Election for Virtual Network Coordination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in enabling multiple smart NICs on a host computer to operate as a single entity, performing virtual network processing and storage operations, while maintaining communication and coordination to handle network management and failover scenarios effectively.
Innovation Solution
Implementing a private communication channel between smart NICs, allowing them to share dynamic state information, configuration data, and pass data messages, and using clustering technology to appear as a single entity to network management systems, with each smart NIC running a smart NIC operating system to perform tasks like virtual switching, routing, and firewall operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple smart NICs operate independently without coordination, then each smart NIC can function autonomously, but they cannot perform virtual network processing as a unified entity
Solution Approach 1:
The patent merges multiple smart NICs into a single logical entity by implementing a leader election mechanism where one smart NIC is designated as the leader to coordinate virtual network processing. The smart NICs share dynamic state information and configuration data through a private communication channel, enabling them to function as a unified team while maintaining individual hardware independence.
2Reliability
If smart NICs share dynamic state information through a private communication channel, then coordination and failover capabilities improve, but communication overhead and system complexity increase
Solution Approach 1:
The patent introduces a private communication channel as an intermediary mechanism between smart NICs. This dedicated channel enables reliable sharing of dynamic state information and configuration data without interfering with external network traffic. The channel supports various topologies (serial, loop, or via physical switch) and provides the foundation for leader election and failover capabilities while isolating coordination traffic from data planes.
3Ease of operation
If a leader election mechanism is implemented among smart NICs, then a single point of contact for network management is established, but additional communication protocols and processing overhead are required
Solution Approach 1:
The patent segments the management functions by designating one smart NIC as the leader responsible for communicating with the network management and control system. The leader election process divides the coordination task among smart NICs, with each participating in the election but only the leader handling external management communications. This segmentation provides a clear single point of contact while distributing the election mechanism's complexity across all participating smart NICs.
4Reliability
If smart NICs use physically separate communication channels, then communication reliability improves, but port availability for networking traffic decreases
Solution Approach 1:
The patent implements multi-functionality in smart NIC ports by enabling them to serve dual purposes: external network communication and internal coordination with other smart NICs. Management ports are specifically designed to handle both external management traffic and private communication channel traffic, eliminating the need for dedicated ports solely for inter-NIC communication. This approach maintains port availability for networking while providing reliable communication channels for coordination.
Data Source
AI summary
Some embodiments provide a method for a first smart NIC of multiple smart NICs of a host computer. Each of the smart NICs is for performing virtual networking operations for a set of data compute machines executing on the host computer. The method determines that the first smart NIC is elected to communicate with a network management and control system that configures the virtual networking operations. The method receives a set of configuration data for the virtual networking operations from the network management and control system. The method provides the received set of configuration data to the other smart NICs of the host computer.