Migrating Stateless Virtual Functions Across Virtual Planes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current PCI and PCIe I/O adapters are integrated into blade servers, limiting their scalability and preventing sharing across blades, as well as not allowing for non-integrated use by multiple blades, which restricts link rate scalability and prevents mechanisms for sharing or migrating virtual functions between system images.
Innovation Solution
A mechanism for natively sharing PCIe adapters between multiple system images across root complexes, enabling unique memory address spaces for each root complex and endpoint, and facilitating communication through shared memory and queuing systems, along with the ability to migrate virtual functions between virtual planes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If PCI and PCIe I/O adapters are integrated into blade servers, then they can be used within the same blade, but they cannot be shared across blades and link rate scalability is limited
Solution Approach 1:
The patent segments the I/O adapter from the blade server integration, allowing the adapter to be a separate, shareable resource. Virtual functions are divided and assigned to different system images, enabling the same physical adapter to serve multiple blades simultaneously without being permanently integrated into any single blade.
Solution Approach 2:
The I/O adapter is designed to perform multiple functions by supporting multiple system images concurrently. The adapter can be dynamically allocated to different blades and system images, making it a universal resource that serves multiple purposes and users rather than being dedicated to a single blade.
2Adaptability or versatility
If I/O adapters are integrated into blades, then they are available to the blade's system image, but they cannot be shared by multiple system images across multiple blades
Solution Approach 1:
The patent introduces a root complex as an intermediary between the I/O adapter and multiple system images. The root complex manages the allocation and routing of I/O requests from different system images to the shared adapter, enabling multi-blade sharing while maintaining reliable resource allocation through centralized control.
Solution Approach 2:
The patent adds a virtualization dimension to the I/O adapter allocation, creating virtual functions that allow the same physical adapter to exist in multiple virtual instances. This dimensional transformation enables the adapter to be simultaneously present and functional for multiple system images across different blades.
3Speed
If link rates are integrated with processor performance, then they scale with processor performance, but current PCI and PCIe implementations do not allow this scalability
Solution Approach 1:
The patent implements dynamic link rate adjustment by allowing the I/O adapter to operate at different speeds depending on the requirements of the connected system images and processors. The link rate can be dynamically scaled up or down based on workload demands and processor performance levels, enabling adaptability rather than fixed integration.
4Adaptability or versatility
If virtual functions are migrated between system images, then resource flexibility improves, but mechanisms for sharing and migrating must be established
Solution Approach 1:
The patent creates virtual copies of the I/O adapter functions through virtual function instances. When migrating workloads, the system transfers the state and configuration between these virtual copies rather than moving physical hardware, enabling flexible workload balancing while keeping the migration mechanism software-based and manageable.
Data Source
AI summary
Mechanisms for migration stateless virtual functions from one virtual plane to another are provided. When a migration of a source virtual function to a destination virtual function in another virtual plane is to be performed, a source single root PCI manager (SR-PCIM) is first interrupted by a multiple root PCI manager (MR-PCIM). Configuration information that defines the source virtual function is then redefined on the destination virtual function for this stateless migration. A function level reset may then be performed on the source virtual function. The destination SR-PCIM may be interrupted by the MR-PCIM with an interrupt for the destination virtual function. A function level reset may then be performed on the destination virtual function. The destination virtual function state may then be changed to an “active” state such that the migrated virtual function begins processing transactions.