Flash SSD Write Policy Selection via Cost Calculation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In log-based data storage devices, repeated write requests to the same logical space lead to increased garbage collection costs, particularly in flash-based SSDs, where random write operations are less efficient than sequential writes, resulting in performance degradation.

Innovation Solution

A method to dynamically determine the write operation policy by calculating and comparing the costs of three write policies: garbage collection, slack space recycling, and in-place updating, to select the policy with the lowest cost, optimizing data storage in flash-based SSDs by considering the sequentiality of storage space.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If repeated write requests are issued on the same logical space, then data storage capacity is utilized, but garbage collection cost rapidly increases

Engineering Contradiction:
Improvedata storage capacity utilizationVSAvoidgarbage collection cost
Core Design Contradiction:
Quantity of substanceVSLoss of energy

Solution Approach 1:

The patent implements dynamic write operation policy selection that adapts to current storage conditions. The system calculates sequential and consecutive write costs for different policies (GC, SSR, IPU) and dynamically selects the policy with the lowest cost, allowing the system to optimize garbage collection efficiency based on real-time utilization patterns and sequentiality metrics

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes operational parameters by selecting different write policies based on calculated costs. When sequentiality is high, the system prefers in-place updating; when sequentiality is low, it prefers garbage collection. This parameter change approach optimizes the balance between storage utilization and garbage collection overhead

Inventive Principle:
Principle #35Parameter changes

2Adaptability or versatility

If random write operations are performed in flash-based SSDs, then write requests are handled flexibly, but throughput is lower compared to sequential writes

Engineering Contradiction:
Improvewrite request handling flexibilityVSAvoidwrite throughput
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The system dynamically adjusts write operation mode based on sequentiality calculation. When consecutive valid pages are detected (high sequentiality), the system performs sequential writes for high throughput. When pages are scattered (low sequentiality), it performs random writes to maintain flexibility. This dynamic adaptation resolves the contradiction between throughput and flexibility

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent applies different write strategies to different regions of storage space based on local sequentiality characteristics. Instead of using a uniform write policy across the entire storage device, the system calculates sequentiality for specific segments and applies appropriate write policies locally, optimizing both throughput and flexibility for each region

Inventive Principle:
Principle #3Local quality

3Reliability

If garbage collection operation is performed frequently, then valid data is maintained, but system performance degrades due to increased overhead

Engineering Contradiction:
Improvedata validity maintenanceVSAvoidsystem performance
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The system implements feedback control by calculating sequential and consecutive write costs before executing write operations. This feedback mechanism allows the system to predict when garbage collection will be expensive and adjust write policies accordingly, reducing unnecessary GC operations and maintaining performance while ensuring data validity

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system performs preliminary cost calculation for different write policies before actually writing data. By evaluating GC cost, SSR cost, and IPU cost in advance, the system selects the optimal policy beforehand, preventing performance degradation from suboptimal write operations and reducing the need for frequent emergency garbage collection

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS9244619B2Method of managing data storage device and data storage device
Publication Date: 2016.01.26 SAMSUNG ELECTRONICS CO LTD
  • US9244619B2 patent drawing
  • US9244619B2 patent drawing
  • US9244619B2 patent drawing

AI summary

A method of managing a data storage device including a memory controller and a memory device includes: calculating a first sequential and consecutive write cost (SCWC) according to a garbage collection (GC) write operation policy, a second SCWC according to a slack space recycling (SSR) write operation policy and a third SCWC according to an in-place updating (IPU) write operation policy respectively, in response to a write request in the memory controller; determining a write operation policy which has a minimum cost of the first through third SCWCs; and writing data in a selected segment in the memory device according to the determined write operation policy.