Network Chip Table Offloading for Scalable Storage Node Connections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage networks face performance degradation due to limited network chip storage space, which cannot accommodate the increasing number of connections between clients and storage devices, leading to inefficient data transmission.
Innovation Solution
The network chip offloads table resources to the storage device's memory, storing connection information there, allowing it to support a larger number of connections by using identifiers to quickly access and manage tables in the storage device's memory.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If connection information is stored in the network chip, then data transmission can be performed efficiently, but the storage space of the network chip is limited and cannot support a large quantity of connections
Solution Approach 1:
The patent extracts the connection information storage function from the network chip and relocates it to the storage device. The network chip retains only the essential data transmission function, while the storage device's memory stores the connection information tables (such as translation tables and queue tables). This separation allows the network chip to support a larger number of connections without being constrained by its limited onboard storage space.
Solution Approach 2:
The patent introduces an intermediary mechanism where the storage device acts as a mediator between the network chip and the connection information. The network chip accesses connection information through memory access requests to the storage device, which contains the required tables. This intermediary approach enables efficient data transmission while utilizing the storage device's larger memory capacity.
2Ease of operation
If the network chip stores connection information for each hard disk and client, then data transmission can be performed, but when the quantity of hard disks or clients increases, the network chip cannot store the information and performance deteriorates
Solution Approach 1:
The patent extracts the connection information storage burden from the network chip and places it in the storage device's memory. The network chip stores only minimal routing information, while detailed connection tables (translation tables mapping connection identifiers to storage device identifiers, and queue tables mapping to memory addresses) are stored in the storage device. This enables the system to handle a large number of hard disks and clients without performance degradation.
Solution Approach 2:
The patent changes the storage dimension from the network chip's limited onboard memory to the storage device's expansive memory space. By relocating connection information storage to another dimension (the storage device's memory space), the system gains substantial capacity to accommodate numerous connections while maintaining fast access through optimized memory structures.
Data Source
AI summary
This disclosure provides methods and storage nodes including network chip and storage devices. In an implementation, a method performed by a network chip comprises generating a command based on an access request from a client, storing the command in a first table in a memory of a storage device connected to the network chip, and executing the command in the first table to access data stored in the storage device.