NAND Program Suspend Control for Random Write SSD Bottlenecks

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

NAND solid-state drives (SSDs) face performance bottlenecks in random write workloads due to program operation latencies, which are an order of magnitude larger than read latencies, exacerbated by the suspension and resumption of program operations to accommodate read operations, leading to increased overhead and reduced throughput.

Innovation Solution

Implementing a dynamic program suspend-resume (PSR) disabling mechanism at the NAND device level, where PSR operations are disabled during random write workloads and enabled only when host reads are present, using a Host Read Counter (HRC) to manage command queues and determine workload type.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If program operation suspend-resume is enabled to accommodate read operations, then read operation latencies are improved, but random write throughput is reduced due to increased program operation overhead

Engineering Contradiction:
Improveread operation latencyVSAvoidrandom write throughput
Core Design Contradiction:
Loss of timeVSProductivity

Solution Approach 1:

The patent applies dynamics by making the program suspend-resume mechanism adaptive and workload-dependent. The controller dynamically determines whether to enable or disable PSR based on the detected workload type (sequential vs. random writes), allowing the system to optimize behavior according to changing conditions rather than using a fixed approach

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes the operational parameter of PSR enablement/disability based on workload characteristics. By detecting workload type and adjusting PSR status accordingly, the system transforms a static feature into a dynamically configured parameter that adapts to maximize performance for different access patterns

Inventive Principle:
Principle #35Parameter changes

2Loss of time

If program operation suspend-resume is used to improve read performance, then host read latencies are reduced, but channel overhead and program operation latencies increase for random write workloads

Engineering Contradiction:
Improvehost read latencyVSAvoidchannel overhead
Core Design Contradiction:
Loss of timeVSLoss of energy

Solution Approach 1:

The patent extracts the PSR mechanism from continuous operation and applies it selectively only when needed. By removing PSR from the default enabled state and activating it only for sequential workloads with pending reads, the system eliminates unnecessary channel overhead and program operation delays in random write scenarios

Inventive Principle:
Principle #2Taking out (Extraction)

3Reliability

If PSR is enabled for all workloads, then read operation performance is maintained, but random write workload performance deteriorates due to unnecessary suspensions

Engineering Contradiction:
Improveread operation performanceVSAvoidrandom write IOPS
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent applies local quality by tailoring PSR behavior to specific workload characteristics rather than applying a uniform approach. The controller analyzes individual workload types and applies PSR only where beneficial (sequential reads with pending programs), while disabling it for workloads where it causes harm (random writes), thus optimizing performance locally for each scenario

Inventive Principle:
Principle #3Local quality

Data Source

PatentUS12487780B2Dynamic program suspend disable for random write SSD workload
Publication Date: 2025.12.02 SK HYNIX NAND PRODUCT SOLUTIONS CORP
  • US12487780B2 patent drawing
  • US12487780B2 patent drawing
  • US12487780B2 patent drawing

AI summary

An embodiment of an electronic apparatus may include one or more substrates, and a controller coupled to the one or more substrates, the controller including circuitry to control access to NAND-based storage media that includes a plurality of NAND devices, determine if a current workload for a particular NAND device of the plurality of NAND devices is a random write workload, and, if so determined, disable a program suspend operation for only the particular NAND device. Other embodiments are disclosed and claimed.