I/O Accelerator Device for Virtualized Storage Data Transfer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional virtualized information handling systems face challenges in achieving high performance for data processing due to limitations in hypervisor driver stacks, leading to low data throughput and high latency when communicating with applications executing on neighboring virtual machines.
Innovation Solution
An I/O accelerator device is introduced, programmed by a storage virtual appliance, which provides managed access to local and remote storage resources using direct memory access (DMA) to improve data throughput and reduce processor workload, and includes additional endpoints for monitoring data transfer operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If hypervisor driver stacks are used for data transfer between virtual machines, then system compatibility and ease of operation are improved, but data throughput decreases and latency increases
Solution Approach 1:
The patent introduces a specialized data transfer interface that acts as an intermediary between virtual machines, bypassing the hypervisor driver stack. This intermediary component enables direct data transfer paths while maintaining compatibility with the existing virtualization architecture, thus resolving the contradiction between ease of operation and data throughput.
Solution Approach 2:
The patent segments the data transfer path by separating the control plane (handled by hypervisor for compatibility) from the data plane (handled by the new interface for high throughput). This segmentation allows different parts of the system to optimize for their respective functions, improving overall productivity without sacrificing ease of operation.
2Ease of operation
If hypervisor driver stacks are used for data transfer between virtual machines, then system compatibility is improved, but latency increases
Solution Approach 1:
The specialized data transfer interface serves as a mediator that reduces latency by providing optimized transfer paths. It maintains system compatibility through standard interfaces while implementing low-latency data transfer mechanisms internally, thus resolving the contradiction between compatibility and latency.
Solution Approach 2:
The patent replaces the software-based hypervisor driver stack with a more efficient data transfer mechanism that operates at lower levels of the system architecture. This substitution reduces processing overhead and latency while maintaining compatibility through standardized interfaces.
3Measurement precision
If processor handles data transfer operations directly, then measurement precision and control are improved, but processor workload increases
Solution Approach 1:
The patent introduces a dedicated data transfer interface as an intermediary that handles the bulk of data transfer operations. This intermediary maintains precise control over data transfers while offloading the workload from the main processor, thus resolving the contradiction between control precision and processor energy consumption.
Solution Approach 2:
The data transfer interface is designed to autonomously manage data transfer operations without requiring constant processor intervention. It implements self-service mechanisms for error handling, flow control, and data routing, maintaining precision while reducing processor workload and energy consumption.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The I/O accelerator device enhances data transfer performance and reduces processor workload by utilizing DMA and additional endpoints for monitoring, thereby improving data throughput and maintaining low latency, even across virtual machines.
Implementation Method 1
An I/O accelerator device is introduced, programmed by a storage virtual appliance, which provides managed access to local and remote storage resources using direct memory access (DMA) to improve data throughput and reduce processor workload
Data Source
AI summary
A method may include receiving one or more monitoring event definitions at an accelerator device from a first logical software entity to a first endpoint of the accelerator device having the first endpoint assigned for access by the first logical software entity, a second endpoint assigned to a second logical software entity such that second endpoint appears to the second logical software entity as a logical hardware adapter, and a third endpoint assigned to a third logical software entity, the accelerator device for accelerating data transfer operations between the second logical software entity and the third logical software entity via the second endpoint and the third endpoint. The method may also include monitoring by the accelerator device for one or more defined monitoring events occurring during the data transfer operations and communicating monitoring information to the first logical software entity from the accelerator device via the first endpoint.


