Virtual I/O Data Segmentation for Cloud Storage Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing methods for virtualization in cloud computing, particularly using SR-IOV, result in high overheads of hardware resources due to the need for complete I/O device storage space for each virtual I/O device, which increases storage and access time.
Innovation Solution
The proposed method involves storing first-type data in memory space within the data processing system, reducing the need for storage space on the I/O device, and storing second-type data on the I/O device for timely access, with the virtual machine monitor managing memory allocation and processing functions to simulate additional operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If complete I/O device storage space is configured for each virtual I/O device, then hardware data can be stored reliably, but hardware resource overheads increase
Solution Approach 1:
The patent segments hardware data into two categories: first-type data (unchanging) stored in host memory and second-type data (changing) stored in I/O device storage space. This segmentation allows the system to maintain reliability for changing data while reducing overall hardware resource overhead by moving static data to host memory.
Solution Approach 2:
The patent changes the storage dimension by moving first-type data from the I/O device storage space to the host machine's memory space. This dimensional shift in data storage location reduces the storage requirements of the I/O device while maintaining data accessibility through the virtualization layer.
2Ease of operation
If all hardware data is stored in I/O device storage space, then data access is straightforward, but data access speed decreases due to frequent I/O device access
Solution Approach 1:
The patent segments data access paths based on data type: first-type data is accessed directly from host memory while second-type data is accessed from I/O device storage space. This segmentation enables faster access for static data while maintaining proper access control for dynamic data.
Solution Approach 2:
The virtual machine monitor acts as an intermediary that manages the differentiated storage architecture. It handles the logic of determining which data type is being accessed and routes access requests appropriately, either to host memory or I/O device storage space, thereby improving overall access speed.
3Adaptability or versatility
If first data is stored in memory space with processing functions, then additional operations can be simulated, but system complexity increases
Solution Approach 1:
The patent copies the functionality of I/O device operations into processing functions stored in host memory. Instead of requiring the physical I/O device to perform operations, software-based processing functions replicate these operations, enabling adaptability while keeping the I/O device itself simpler.
Solution Approach 2:
The patent substitutes mechanical/hardware operations with software-based processing functions. The processing functions in host memory replace the need for complex hardware logic in the I/O device, achieving operational versatility through software rather than hardware complexity.
Data Source
Figure 1~3
Figure 4
Figure 5
AI summary
This application provides a data processing method, apparatus, and device. The method includes: In a data storage process, a virtual machine monitor stores, in first memory space of a server, first-type data that is in hardware data of a virtual I/O device. In a data access process, after receiving an I/O access request, a virtual machine identifies a type of hardware data that the I/O access request requests to access. When identifying that the type of the hardware data that the I/O access request requests to access is first-type data, the virtual machine accesses the data in the first memory space of the server. This reduces hardware resource usage of the I/O device.