SSD Block Reclamation Scoring for Write Amplification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In solid-state storage media, data blocks can only be erased and programmed a limited number of times before transistors degrade, leading to reduced longevity, and existing reclamation methods like garbage collection increase write amplification, which further reduces media longevity and bandwidth.

Innovation Solution

A method and apparatus for selecting storage blocks for reclamation based on operational effects such as write amplification and quality-of-service criteria, using a scoring system to identify blocks that minimize write amplification and maximize reuse, while maintaining quality-of-service levels.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If garbage collection is performed to reclaim storage blocks, then storage capacity is improved, but write amplification increases and media longevity decreases

Engineering Contradiction:
Improvestorage capacityVSAvoidmedia longevity
Core Design Contradiction:
Quantity of substanceVSDuration of action of stationary object

Solution Approach 1:

The patent changes the selection parameters for garbage collection by introducing a scoring mechanism that evaluates multiple factors including write amplification impact, cache hit ratio, and data age. This transforms the reclamation process from simple FIFO or random selection to an optimized selection based on calculated scores that balance capacity reclamation with media longevity preservation.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent performs preliminary evaluation and scoring of storage blocks before reclamation. By pre-calculating scores that predict the impact of reclamation on write amplification and cache performance, the system can select blocks that minimize negative effects before actually performing the reclamation operation.

Inventive Principle:
Principle #10Preliminary action

2Quantity of substance

If storage blocks are reclaimed using traditional garbage collection, then available storage space increases, but bandwidth and performance decrease due to increased write amplification

Engineering Contradiction:
Improveavailable storage spaceVSAvoidbandwidth
Core Design Contradiction:
Quantity of substanceVSProductivity

Solution Approach 1:

The patent modifies the garbage collection parameters by introducing a scoring system that explicitly considers bandwidth impact through cache hit ratio calculations. Blocks with higher scores indicate lower expected write amplification and better cache performance, thus preserving bandwidth during reclamation operations.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent incorporates feedback mechanisms by monitoring cache hit ratios and using this information to adjust block selection scores. The system continuously evaluates the impact of reclamation decisions on cache performance and uses this feedback to optimize future selections, thereby maintaining bandwidth during capacity management.

Inventive Principle:
Principle #23Feedback

3Quantity of substance

If frequent reclamation operations are performed to maintain storage capacity, then storage availability is improved, but cell degradation accelerates due to repeated erase cycles

Engineering Contradiction:
Improvestorage availabilityVSAvoidcell longevity
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The patent changes the reclamation frequency and selection parameters by incorporating cell age and erase cycle history into the scoring mechanism. This allows the system to identify blocks that are safe to reclaim without accelerating cell degradation, balancing storage availability with reliability concerns.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent performs preliminary assessment of block reclamation impact on cell longevity before executing reclamation. By evaluating factors such as data age, access patterns, and cell wear history in advance, the system can schedule reclamation operations that maintain storage availability while minimizing additional stress on storage cells.

Inventive Principle:
Principle #10Preliminary action

4Loss of energy

If selective reclamation based on scoring is implemented, then write amplification is reduced, but computational complexity and processing time increase

Engineering Contradiction:
Improvewrite amplificationVSAvoidcomputational complexity
Core Design Contradiction:
Loss of energyVSDevice complexity

Solution Approach 1:

The patent manages computational complexity by defining a structured scoring mechanism with specific parameters (write amplification factor, cache hit ratio, data age) that can be calculated efficiently. The scoring system transforms complex multi-factor evaluation into a manageable computational framework that reduces overall system complexity while achieving the goal of minimizing write amplification.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS9146688B2Advanced groomer for storage array
Publication Date: 2015.09.29 SANDISK TECHNOLOGIES LLC
  • US9146688B2 patent drawing
  • US9146688B2 patent drawing
  • US9146688B2 patent drawing

AI summary

Techniques are disclosed relating to reclaiming data on recording media. In one embodiment, an apparatus has a solid-state memory array including a plurality of blocks. The solid-state memory array may implement a cache for one or more storage devices. Respective operational effects are determined relating to reclaiming ones of the plurality of blocks. One of the plurality of blocks is selected as a candidate for reclamation based on the determined operational effects, and the selected block is reclaimed. In some embodiments, the determined operational effects for a given block indicate a number of write operations to be performed to reclaim the given block. In some embodiments, operational effects are determined based on criteria relating to assigned quality-of-service levels. In some embodiments, operational effects are determined based on information relating to virtual storage units.