Nested Hypervisor NIC Pass-Through via Cascaded Extenders

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current technologies cannot pass a network interface card (NIC) through multiple nested hypervisors, leading to network operations challenges and performance overhead due to virtual network stacking inconsistencies and emulated virtual NIC device model variations, making it difficult to maintain high and predictable networking performance during VM migration.

Innovation Solution

The use of extender components like VM-FEX or IEEE 802.1BR Port Extender, along with a delegation model API, allows for transparent bypass of multiple nested hypervisors by cascading extender components and propagating pass-through operations between hypervisors, enabling direct network communication between the guest OS and physical NIC without hypervisor intervention.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If a network interface card is passed through a single hypervisor, then network performance is improved, but passing it through multiple nested hypervisors creates virtual network stacking inconsistencies and device model variations

Engineering Contradiction:

Improvenetworking performanceVSAvoidvirtual network stacking consistency

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent segments the network pass-through operation into distinct phases: emulation mode for initial operation and pass-through mode for high performance. The extender component is also segmented across multiple hypervisor levels, with each hypervisor having its own extender instance that coordinates with others to achieve consistent behavior through the delegation model API.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The extender component acts as an intermediary between the virtual network stack and physical NIC. In multi-hypervisor scenarios, extenders at different hypervisor levels serve as mediators that coordinate through the delegation model API, allowing upper-level extenders to delegate operations to lower-level extenders, thereby maintaining consistency across the nested architecture.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Adaptability or versatility

If network operations go through multiple nested hypervisors, then VM migration flexibility is improved, but performance overhead increases due to software intervention

Engineering Contradiction:

ImproveVM migration flexibilityVSAvoidnetworking throughput

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The system dynamically switches between emulation mode and pass-through mode based on operational requirements. During VM migration operations, the system can operate in emulation mode for compatibility, while during high-performance network operations, it switches to pass-through mode to eliminate software overhead. This dynamic behavior is coordinated across multiple hypervisors through the extender delegation model.

Inventive Principle:

Principle #15Dynamics

3Adaptability or versatility

If emulated virtual NIC device models are used across nested hypervisors, then compatibility is improved, but performance predictability deteriorates

Engineering Contradiction:

Improvedevice model compatibilityVSAvoidnetworking performance predictability

Core Design Contradiction:

Adaptability or versatilityVSManufacturing precision

Solution Approach 1:

The extender component implements self-service by automatically detecting the operational mode requirements and coordinating with extenders at other hypervisor levels through the delegation model API. This self-coordination ensures that all extenders in the nested hierarchy maintain consistent state without requiring manual configuration, thereby ensuring both compatibility and predictable performance.

Inventive Principle:

Principle #25Self-service

4Productivity

If pass-through mode is implemented in multiple nested hypervisors, then network performance is improved, but system complexity increases due to coordination requirements

Engineering Contradiction:

Improvenetworking throughputVSAvoidextender coordination complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent applies the nested doll principle by placing extender components at multiple nested hypervisor levels, with each extender encapsulating the functionality needed at its specific level. The delegation model API enables outer-level extenders to delegate operations to inner-level extenders, creating a hierarchical coordination structure that manages complexity through clear delegation relationships while maintaining high performance.

Inventive Principle:

Principle #7Nested doll (Nesting)

Data Source

PatentUS9176767B2Network interface card device pass-through with multiple nested hypervisors

Publication Date: 2015.11.03 CISCO TECHNOLOGY INC

AI summary

In a data center computing system, multiple nested hypervisors are run, including an outer hypervisor and at least one inner hypervisor running as a virtual machine on top of the outer hypervisor. A guest operating system is run as a virtual machine in the innermost hypervisor. An emulated network interface card device is executed in all hypervisors. An extender component is executed in the outer hypervisor and an extender component is executed in the inner hypervisors such that the extender components in the outer hypervisor and in the inner hypervisors are architecturally cascaded. An interface for the guest operating system is assigned to the emulated network interface card device in each of the outer hypervisor and the inner hypervisors to enable network communications to bypass the outer hypervisor and the inner hypervisors.