RDMA Log Replication Model for Storage I/O Latency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Cloud computing environments face inefficiencies in storage I/O operations due to traditional data transfer methods, which involve multiple copies and context switches, leading to increased latency and CPU overhead.

Innovation Solution

The implementation of a Remote Direct Memory Access (RDMA) replication model, which uses RDMA write and send operations to directly transfer data and metadata between hosts, eliminating the need for CPU intervention and reducing latency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If traditional data transfer methods are used for storage I/O operations, then data can be transferred between systems, but multiple copies and context switches occur leading to increased latency and CPU overhead

Engineering Contradiction:
ImprovelatencyVSAvoiddata transfer process complexity
Core Design Contradiction:
Loss of timeVSDevice complexity

Solution Approach 1:

The patent extracts the data transfer function from the CPU and places it directly in the memory subsystem through RDMA. The CPU is removed from the data transfer path, allowing memory-to-memory transfers without CPU intervention, thereby reducing latency and CPU overhead while simplifying the overall system architecture.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces a memory intermediary that enables direct data transfer between remote memory spaces. This intermediary mechanism (RDMA) eliminates the need for traditional CPU-mediated data copying and context switching, reducing the complexity of the data transfer process while maintaining reliable data communication.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If traditional data transfer methods are used, then data can be replicated to remote systems, but multiple copies and context switches increase CPU overhead

Engineering Contradiction:
ImprovethroughputVSAvoidCPU involvement
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent extracts the data replication function from CPU control and implements it through direct memory access. By removing CPU involvement from the replication process, the system achieves higher throughput with reduced CPU overhead, as data can be replicated directly between memory spaces without CPU intervention.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent enables the memory subsystem to perform data replication autonomously through RDMA. The memory spaces can directly transfer data to remote locations without requiring CPU service or intervention, thereby increasing productivity while reducing the burden on the CPU.

Inventive Principle:
Principle #25Self-service

3Adaptability or versatility

If data and metadata are transferred separately, then transfer flexibility is improved, but additional operations and potential inconsistencies may occur

Engineering Contradiction:
Improvetransfer flexibilityVSAvoiddata consistency
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The patent segments the data transfer operation into separate data portion and metadata portion transfers. This segmentation allows flexible independent transfer of each component while maintaining consistency through coordinated RDMA operations. The separate transfers enable adaptability to different storage scenarios while ensuring reliability through proper ordering and completion tracking.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS20250159043A1Remote direct memory access data replication model
Publication Date: 2025.05.15 MICROSOFT TECHNOLOGY LICENSING LLC
  • US20250159043A1 patent drawing
  • US20250159043A1 patent drawing
  • US20250159043A1 patent drawing

AI summary

A computer-implemented method for replicating a log to a remote computer system is disclosed. The method involves identifying a log comprising a data portion and a metadata portion for replication. The data portion is sent to the remote computer system using a Remote Direct Memory Access (RDMA) write operation, while the metadata portion is sent using a first RDMA send operation after the data portion has been sent. The method further includes identifying a second RDMA send operation received from the remote computer system, which indicates the completion of the first RDMA send operation. Based on identifying the second RDMA send operation, the method determines the completion of log replication to the remote computer system. This method enables efficient and reliable replication of logs in a computer system.