IO Virtualization System Resource Allocation via PCIe Bridge
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional network architectures face limitations in sharing PCIe devices across multiple host systems, requiring proprietary solutions and extensive infrastructure changes, and lack centralized management and transparent PCIe connections.
Innovation Solution
The Input/Output Virtualization (IOV) system enables sharing of PCIe devices by using a scalable, low-latency transport fabric to connect host computers and IO devices, with a management CPU managing resources and providing transparent PCIe connections, allowing multifunction devices to be shared across hosts and maintaining reliability and quality of service.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If proprietary IO resource modules or drivers are used to implement IO virtualization, then IO device sharing capability is improved, but device complexity and infrastructure requirements increase
Solution Approach 1:
The patent introduces a PCIe bridge as an intermediary component that enables IO device sharing between multiple hosts without requiring proprietary IO resource modules or drivers. The PCIe bridge translates and manages IO requests from multiple hosts to shared IO devices, providing a standard-based solution that maintains compatibility with existing PCIe infrastructure while enabling virtualization capabilities.
2Adaptability or versatility
If FCoE is implemented to transport storage traffic, then network flexibility is improved, but infrastructure replacement requirements increase
Solution Approach 1:
The patent segments the IO virtualization function into the PCIe bridge component, which handles protocol translation and resource management, while allowing existing Ethernet infrastructure to remain unchanged. This segmentation enables FCoE-like functionality for storage traffic without requiring replacement of the entire switching fabric or IO devices, as the PCIe bridge performs the necessary protocol conversions at the host level.
3Reliability
If ExpressEther is used to transport PCIe traffic, then connection transparency is improved, but device sharing capability deteriorates
Solution Approach 1:
The patent merges the functionality of ExpressEther (transparent PCIe transport) with IO virtualization capabilities in the PCIe bridge. The bridge maintains transparent PCIe connections for each host while simultaneously enabling device sharing by intercepting and redirecting IO requests to shared IO devices, thus combining both transparency and sharing capabilities in a single component.
4Adaptability or versatility
If multiple root complexes are used in PCIe hierarchy, then system scalability is improved, but implementation complexity increases
Solution Approach 1:
The PCIe bridge acts as an intermediary that enables multi-root IOV functionality without requiring complex fabric and endpoint implementations. By placing the virtualization logic in the PCIe bridge, the system achieves multi-root capability with simplified architecture, as the bridge manages resource allocation and request routing between multiple hosts and shared IO devices without requiring complex coordination at the fabric or endpoint levels.
Data Source
AI summary
An Input/Output (IO) Virtualization (IOV) system provides a mechanism for sharing computer peripheral devices or resources between multiple host computers by presenting a single resource multiple times to numerous host systems. The IOV system provides IO virtualization and host-to-host communication services to the host computers. The system comprises device interfaces coupled to the IO devices. Each IO device comprises at least one endpoint function (EPF). Host interfaces are each coupled to one of a number of host computers. Each host interface includes a proxy, and each proxy comprises configuration data.