DMA Controller Intermediate Memory Reduces PCIe Latency

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In Hyper-Converged infrastructure computer systems, accessing storage apparatuses across nodes increases latency due to protocol conversion and CPU access requests, leading to reduced total I/O performance.

Innovation Solution

A computer system with a communication network connecting nodes, featuring a storage device and a communication device with a controller and intermediate memory for data transmission, optimizing data exchange between nodes through PCIe cards with integrated Storage Class Memory (SCM) for caching and DMA transfers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a node accesses storage apparatuses in other nodes through the communication network, then data sharing and system scalability are improved, but latency increases due to protocol conversion and CPU access requests

Engineering Contradiction:

Improvedata sharing capabilityVSAvoidaccess latency

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent introduces a communication device as an intermediary component that includes an intermediate memory for caching data and a DMA controller for direct memory access. This intermediary enables nodes to access storage apparatuses in other nodes without requiring CPU intervention or protocol conversion for every access, thereby reducing latency while maintaining data sharing capability across the network

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent implements preliminary action by pre-fetching and caching data in the intermediate memory before actual access is needed. The DMA controller is pre-configured with transfer parameters, and data is prepared in advance in the intermediate memory, allowing nodes to access stored data quickly without real-time protocol conversion or CPU processing delays

Inventive Principle:

Principle #10Preliminary action

2Adaptability or versatility

If data is transmitted between nodes through the communication network, then system scalability and distributed storage are improved, but total I/O performance decreases due to increased latency

Engineering Contradiction:

Improvesystem scalabilityVSAvoidtotal I/O performance

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The communication device serves as a mediator that buffers and manages data transmission between nodes. The intermediate memory acts as a local cache that reduces the need for repeated network transmissions, while the DMA controller efficiently manages data transfer without CPU involvement, thereby maintaining high I/O performance in scalable distributed systems

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent uses copying by creating local copies of frequently accessed data in the intermediate memory of communication devices. This allows nodes to access copied data locally rather than repeatedly transmitting over the network, significantly improving I/O performance while maintaining system scalability

Inventive Principle:

Principle #26Copying

Data Source

PatentUS10733118B2Computer system, communication device, and storage control method with DMA transfer of data

Publication Date: 2020.08.04 HITACHI VANTARA LTD

AI summary

This computer system is configured by connecting a plurality of computers via a communication network. At least one computer among the computers has a storage device and a communication device. The communication device has: a controller that controls data transmission/reception via the communication network; and an intermediate memory that stores data transmitted/received between the storage device and other calculators on the communication network.