Distributed Storage Controller with Peer-to-Peer SSD Communication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Disaggregated storage architectures fail to effectively utilize I/O bandwidth, processing, and buffer memory of solid-state drives (SSDs) and face challenges in scaling due to centralized compute resources, leading to inefficiencies in read/write operations and reliability in the face of storage system scaling, drive failures, and power interruptions.
Innovation Solution
A distributed storage system with a storage controller and multiple SSDs that utilize a service mapping module, media manager, and logical map update module to identify and manage target storage devices for data requests, distribute compute resources, and ensure reliable read/write operations through distributed processing and logical mapping across multiple SSDs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If centralized compute resources are placed in a storage controller, then coordination and control of storage operations is improved, but scalability is worsened as the number and size of SSDs increases
Solution Approach 1:
The patent divides the centralized storage controller into distributed storage devices, where each storage device has its own processor and buffer memory. This segmentation allows each device to independently handle I/O operations and data caching, eliminating the scalability bottleneck of centralized control while maintaining coordinated operation through standardized protocols.
Solution Approach 2:
The patent introduces a new architectural dimension by implementing a peer-to-peer communication mechanism between storage devices, allowing them to directly exchange data and coordination information without passing through a centralized controller. This dimensional shift from hierarchical to distributed topology enables linear scalability.
2Adaptability or versatility
If disaggregated storage architecture is implemented, then flexibility and cost savings are improved, but utilization of I/O bandwidth, processing, and buffer memory of SSDs is worsened
Solution Approach 1:
Each storage device in the disaggregated architecture is equipped with its own processor and buffer memory, enabling it to autonomously perform data caching, I/O operation handling, and protocol translation. This self-service capability ensures that buffer memory and processing resources are fully utilized locally, eliminating the underutilization problem while preserving architectural flexibility.
3Adaptability or versatility
If distributed processing across storage devices is implemented, then scalability and reliability are improved, but system complexity is worsened
Solution Approach 1:
The patent implements a universal interface and standardized communication protocol that enables each storage device to perform multiple functions including data storage, caching, I/O handling, and inter-device communication. This multi-functionality reduces system complexity by eliminating the need for specialized components while maintaining distributed processing capabilities.
Data Source
AI summary
Example multi-device storage systems and methods provide disaggregated read/write operations. Read and/or write data requests are received from a host system at a storage controller for a plurality of storage devices. A target service host storage device is identified that hosts a target portion of a logical map of storage locations distributed among the plurality of storage devices. The target service host storage device identifies a destination storage device corresponding to a target location for the request. A data transfer from the host system to the destination storage device is processed for the request and the logical map of the target service host device is updated. The target service host storage device and the destination storage device may be different storage devices among the plurality of storage devices.