Hardware Switch for Hypervisor Guest OS Intra-Server Ethernet Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Ethernet switches cannot loop packets back onto the same port, preventing effective switching between guest operating systems sharing a common physical network interface card (pNIC), leading to CPU utilization penalties and inefficient memory bandwidth usage in hypervisor systems, especially in multi-root I/O virtualization environments.
Innovation Solution
A hardware switch embedded in a pNIC with PCIe interface, transmit and receive queues, and routing logic that performs filtering and lookups based on MAC addresses and VLANs, allowing for intra-server and inter-server switching between guest OSs without relying on the hypervisor, by forwarding Ethernet data between PCIe bus and 10 GbE physical ports.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a virtual switch (vSwitch) is employed in the hypervisor to connect vNICs and perform switching between guest OSs, then switching functionality is achieved, but CPU utilization increases due to hypervisor intervention for every I/O operation
Solution Approach 1:
The patent extracts the switching functionality from the hypervisor software layer and places it directly in the pNIC hardware. The pNIC now contains an integrated hardware switch that can perform Ethernet switching operations independently, removing the need for hypervisor intervention in packet forwarding decisions. This extraction of the switching function from software to hardware directly reduces CPU utilization while maintaining full switching capability between vNICs.
Solution Approach 2:
The patent introduces a new intermediary component - the hardware switch embedded in the pNIC - that acts as a mediator between vNICs and the external network. This hardware switch receives Ethernet frames from vNICs, performs switching decisions based on destination MAC addresses, and forwards frames to the appropriate destination without requiring hypervisor processing. The hardware switch serves as an intelligent intermediary that handles switching operations autonomously.
2Productivity
If data is copied between virtual memory spaces assigned to source and destination OSs for routing, then packet routing is achieved, but memory bandwidth is wasted due to the copying step
Solution Approach 1:
The patent replaces the mechanical copying process with a more efficient mechanism. Instead of copying data from source OS virtual memory to destination OS virtual memory through the hypervisor, the hardware switch in the pNIC performs direct frame forwarding at the Ethernet level. The hardware switch operates on Ethernet frames before they are fully processed into higher-layer protocols, enabling direct memory-to-memory transfers between guest OSs without intermediate copying steps, thus preserving memory bandwidth.
3Ease of operation
If guest OSs communicate through the normal networking stack via a common pNIC, then network communication is achieved, but switching between guest OSs cannot be performed by the external Ethernet switch
Solution Approach 1:
The patent merges the Ethernet switching functionality directly into the pNIC hardware, combining the network interface card and switch into a single integrated device. This merging allows the pNIC to handle both external network communication and internal switching between guest OSs simultaneously. The hardware switch within the pNIC can inspect Ethernet frames and perform switching decisions based on destination MAC addresses, enabling external switches to properly handle traffic between guest OSs while maintaining standard Ethernet communication protocols.
Data Source
AI summary
A hardware switch for use with hypervisors and blade servers is disclosed. The hardware switch enables switching to occur between different guest OSs running in the same server, or between different servers in a multi-root IOV system, or between different guest OSs running in the same server in single-root IOV systems. Whether embedded in a host bus adapter (HBA), converged network adapter (CNA), network interface card (NIC) or other similar device, the hardware switch can provide fast switching with access to and sharing of at least one external network port such as a Fibre Channel (FC) port, 10 Gigabit Ethernet (10 GbE) port, FC over Ethernet (FCoE) port, or other similar port. The hardware switch can be utilized when no hypervisor is present or when one or more servers have hypervisors, because it allows for switching (e.g. Ethernet switching) between the OSs on a single hypervisor.