Ethernet Switch RAID Control for NVMe Storage Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing NVMe over Fabrics SSD configurations face challenges in providing erasure code data protection due to increased latency and cost associated with external RAID controllers, and the need for improved cost-performance optimization in Ethernet-attached SSDs.
Innovation Solution
A system and method utilizing an Ethernet switch with a RAID controller or state machine, incorporating a low-latency cache and local central processing unit to manage and protect data across an array of SSDs, optimizing data storage by reducing latency through a cache device with latency five times lower than SSDs and integrating RAID control circuits within the Ethernet switch.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If an external RAID controller is used to provide erasure code data protection, then data protection capability is improved, but system latency increases and performance degrades
Solution Approach 1:
The patent extracts the RAID control functionality from external controllers and integrates it directly into the NVMe SSD devices. Each SSD contains a RAID control circuit that can locally execute erasure code operations, eliminating the need for external RAID controllers and reducing system latency while maintaining data protection capabilities.
Solution Approach 2:
The patent introduces a cache memory as an intermediary between the NVMe SSDs and the host system. The cache temporarily stores data and erasure code information, allowing rapid local processing of RAID operations without requiring constant communication with external controllers, thus reducing latency while maintaining reliability.
2Reliability
If multiple dual pathing IO SSDs are used to improve system availability, then fault protection is improved, but system cost increases and performance slightly degrades
Solution Approach 1:
The patent merges the RAID control functionality into each individual NVMe SSD device, creating a distributed RAID architecture. This eliminates the need for complex external RAID controllers and multiple dual-pathing interfaces, reducing system cost and complexity while maintaining fault protection through the inherent redundancy of erasure coding across multiple SSDs.
Solution Approach 2:
Each NVMe SSD becomes self-sufficient with integrated RAID control circuits that can autonomously perform erasure code operations. This self-service capability allows each device to contribute to the overall fault protection without requiring complex coordination or additional hardware paths, simplifying the system architecture while maintaining high availability.
Data Source
AI summary
A system and method for providing erasure code data protection for an array of solid state drives. The solid state drives are connected to an Ethernet switch which includes a RAID control circuit, or a state machine, to process read or write commands that may be received from a remote host. The RAID control circuit, if present, uses a low-latency cache to execute write commands, and the state machine, if present, uses a local central processing unit, which in turn uses a memory as a low-latency cache, to similar effect.