Parallel Data Transmission for Storage Node Replication Latency

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In cluster storage platforms, efficient low-latency communication between nodes is crucial for performance and reliability, but existing systems often delay data transmission to backup nodes, increasing latency due to the need for the owner node to process requests before forwarding data to the backup.

Innovation Solution

An initiating node sends service requests or change data to both the owner and backup nodes in parallel, allowing the backup node to receive data before the owner has completed processing, reducing the need for large data transfers and minimizing latency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the owner node processes storage requests sequentially before forwarding data to the backup node, then data integrity and consistency are ensured, but transmission latency increases and processing time is delayed

Engineering Contradiction:

Improvedata consistencyVSAvoidtransmission latency

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The initiating node performs preliminary action by sending data to the backup node before the owner node completes processing. The initiating node receives the storage request, identifies the backup node, and initiates data transmission to the backup node concurrently with the owner node's processing, rather than waiting for processing completion. This preliminary action reduces latency while maintaining consistency through proper coordination protocols.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system transitions from a sequential one-dimensional processing approach to a parallel multi-dimensional approach. Instead of processing requests sequentially in a single timeline, the initiating node operates in multiple dimensions by simultaneously coordinating with both the owner node and backup node, enabling overlapping operations that reduce overall latency while ensuring data consistency through coordinated protocol.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

2Ease of operation

If the initiating node waits for the owner node to complete processing before sending data to the backup node, then processing simplicity is maintained, but storage operation speed decreases

Engineering Contradiction:

Improveprocessing simplicityVSAvoidstorage operation speed

Core Design Contradiction:

Ease of operationVSProductivity

Solution Approach 1:

The initiating node performs preliminary action by proactively sending data to the backup node before the owner node completes processing. This eliminates the need to wait for sequential completion, significantly improving storage operation speed. The initiating node receives the storage request, identifies the backup node, and initiates data transmission concurrently with the owner node's processing, thereby maintaining simplicity while enhancing productivity.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements continuity of useful action by ensuring that data transmission to the backup node occurs continuously and concurrently with the owner node's processing rather than sequentially. The initiating node maintains continuous engagement with both nodes, overlapping transmission and processing operations to maximize system throughput and operational speed.

Inventive Principle:

Principle #20Continuity of useful action

3Device complexity

If data transmission to backup node is delayed until owner processing completes, then coordination complexity is reduced, but replication latency increases

Engineering Contradiction:

Improvecoordination complexityVSAvoidreplication latency

Core Design Contradiction:

Device complexityVSLoss of time

Solution Approach 1:

The initiating node performs preliminary action by sending data to the backup node before the owner node completes processing. This preliminary transmission action reduces replication latency significantly. The initiating node receives the storage request, identifies the backup node, and initiates data transmission to the backup node concurrently with the owner node's processing, rather than waiting for processing completion.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system employs feedback mechanisms where the initiating node receives confirmation from both the owner node and backup node about the storage operation status. This feedback allows the initiating node to coordinate the transmission timing optimally, ensuring data consistency while minimizing replication latency through intelligent feedback-driven synchronization.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20250013662A1Low latency communications for nodes in replication relationships

Publication Date: 2025.01.09 NVIDIA CORP

AI summary

An initiating node (C) in a storage platform (100) receives a modification request (312, 314) for changing an object (0). The initiating node (C), using system configuration information (127), identifies an owner node (A) and a backup node (B) for the object (0) and sends change data (324, 334) to the owner node (A) and the backup node (B). The owner node (A) modifies the object (0) with the data (324) from the initiating node (C) and sends an update request (352) that does not include the data (324) to the backup node (B). The backup node (B) modifies a backup object (0′) with data (334) from the initiating node (C).