SSD Peer Group Data Protection Offload
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data protection methods for storage systems, such as RAID configurations, face limitations in performance and cost, with software RAID impacting system performance and hardware RAID being costly, while also requiring intermediary devices and limited by hardware performance.
Innovation Solution
A storage system utilizing a peer group of solid-state drives (SSDs) with host and peer-interface ports connected through networks like PCIe, Ethernet, or Fibre Channel, allowing for coordinated data-protection computations and offloading CPU-intensive tasks without performance tradeoffs, enabling direct communication between host CPUs and SSDs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If software RAID is used, then acquisition cost is low and ease of upgrade is high, but system performance is adversely impacted due to sharing CPU resources
Solution Approach 1:
The patent extracts the RAID computation functionality from the host CPU and relocates it to dedicated RAID controller hardware. This separation allows the host CPU to focus on general-purpose computing while the RAID controller handles data protection computations independently, resolving the performance penalty imposed by software RAID without sacrificing the cost benefits of avoiding expensive hardware RAID controllers.
Solution Approach 2:
The patent introduces a dedicated RAID controller as an intermediary component between the host system and storage devices. This intermediary handles all RAID computations, data protection operations, and storage management tasks, freeing the host CPU from these responsibilities while maintaining system integration and avoiding the need for expensive proprietary hardware solutions.
2Productivity
If hardware RAID is used, then RAID computation is offloaded from host CPU and performance is high, but acquisition cost is high due to additional hardware components
Solution Approach 1:
The patent employs a cost-effective RAID controller implementation that avoids expensive proprietary hardware components while maintaining high performance. The controller uses standard, readily available hardware components and open-standard interfaces, providing a disposable yet high-performance solution that eliminates the need for expensive, vendor-locked hardware RAID controllers.
3Adaptability or versatility
If hardware RAID controller is used, then complex RAID configurations are supported, but device complexity increases with additional hardware components
Solution Approach 1:
The patent implements a universal RAID controller that supports multiple RAID configurations (RAID 0, 1, 5, 6, 10, and other custom configurations) through a single hardware platform. The controller uses standardized interfaces and modular design, allowing it to adapt to different storage requirements without requiring additional specialized hardware components for each RAID level.
Data Source
AI summary
A storage system includes a plurality of solid-state drives (SSDs) in which each SSD is an endpoint of a peer group in which one SSD is a primary SSD of the peer group, or all SSDs are treated as equal peers. Each SSD further includes a host interface port coupled to a host computing device that receives receive input/output (I/O) communications from the host computing device, a peer-interface port coupled to the peer-interface port of each other SSD in the peer group; and a controller coupled to the host interface port and the peer-interface port. The primary SSD controller is responsive to I/O communications received from the host computing device to provide data-protection computations relating to a coordinated data-protection configuration provided by the peer group and to pass the coordinated data-protection configuration information to the controllers of the other end points of the peer group through the peer-interface port.