Distance-Based SSD Bandwidth Tiers for NAND Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional SSDs are limited by the slowest speed in the chain, typically corresponding to the NAND device with the longest physical channel to the controller, leading to inefficient performance and the need for duplicate storage nodes to meet both latency and density requirements, increasing power, cost, and space demands.
Innovation Solution
Organize NAND devices into multiple physical clusters based on their distance from the controller and inherent performance, optimizing bus speeds to match each cluster's characteristics, allowing higher performance NAND devices closer to the controller and higher density devices further away, and presenting these clusters as separate drives to manage data placement and access requests.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the SSD uses a single uniform bus speed for all NAND devices, then the system is simple to manage, but the overall performance is limited by the slowest device
Solution Approach 1:
The patent segments the SSD storage system into multiple performance tiers (first tier and second tier) based on NAND device characteristics and physical location. Each tier operates with its own optimized bus speed, allowing fast devices to run at higher speeds without being constrained by slower devices. This segmentation resolves the contradiction by enabling high performance while maintaining manageable complexity through structured organization.
Solution Approach 2:
The patent applies local quality by assigning different bus speeds to different physical locations and device types within the SSD. Devices closer to the controller operate at faster bus speeds, while devices further away operate at slower speeds. This localized optimization allows each region to operate at its optimal performance level, resolving the contradiction between overall performance and management complexity.
2Adaptability or versatility
If the SSD stores both high performance and high density NAND devices, then capacity and speed requirements are met, but duplicate storage nodes are needed increasing power and cost
Solution Approach 1:
The patent merges previously separate high-performance and high-density storage functions into a single SSD device by organizing different NAND device types into different performance tiers within the same system. This consolidation eliminates the need for duplicate storage nodes, reducing power consumption while maintaining both performance and capacity capabilities in one integrated device.
Solution Approach 2:
The patent makes the single SSD device universal by enabling it to perform both high-performance and high-density storage functions simultaneously through its multi-tier architecture. The SSD can handle various workloads across different tiers, providing adaptability for different storage requirements without needing separate specialized devices, thus reducing overall system power consumption.
3Quantity of substance
If high performance NAND devices are placed further from the controller, then density is improved, but access latency increases
Solution Approach 1:
The patent applies local quality by matching device characteristics to their physical locations within the SSD. High-performance NAND devices are positioned closer to the controller where they can operate at lower latency, while high-density devices are placed further away where their lower speed is acceptable. This spatial optimization resolves the contradiction by ensuring each location's performance characteristics match its operational requirements.
Data Source
AI summary
An embodiment of an electronic apparatus may include a substrate and a controller coupled to the substrate, the controller including circuitry to control access to a NAND-based storage media that includes a plurality of NAND devices located on the substrate and organized into two or more physical clusters with each NAND device uniquely assigned to one of the two or more physical clusters, perform data access to a first physical cluster of the two or more physical clusters at a first bandwidth, and perform data access to a second physical cluster of the two or more physical clusters at a second bandwidth that is slower than the first bandwidth. Other embodiments are disclosed and claimed.