Adaptive Multi-Stream Classification for SSD Performance

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Multi-stream SSDs face challenges in managing numerous streams specified by host devices, leading to increased resource consumption, garbage collection activities, and reduced performance due to inefficient stream management and lack of access to internal storage metrics.

Innovation Solution

A data storage device dynamically classifies streams into stream sets based on internally monitored performance metrics, treating streams with similar metrics as a single stream for write operations, and determines an efficiency score to select the most efficient stream classification for operations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If the data storage device manages numerous streams specified by host devices, then the device can handle diverse data access patterns, but resource consumption increases and performance decreases

Engineering Contradiction:

Improvestream handling capabilityVSAvoiddevice performance

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent merges multiple streams with similar performance metrics into a single stream set, reducing the total number of streams the device must manage. This combining approach maintains the ability to handle diverse access patterns while decreasing resource consumption and improving overall device performance.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent creates stream sets that can represent multiple original streams, giving each stream set the universal capability to handle various data access patterns that were previously distributed across individual streams. This multi-functionality reduces the number of managed entities while preserving adaptability.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Adaptability or versatility

If the data storage device manages numerous streams specified by host devices, then the device can handle diverse data access patterns, but garbage collection activities increase

Engineering Contradiction:

Improvestream handling capabilityVSAvoidgarbage collection overhead

Core Design Contradiction:

Adaptability or versatilityVSLoss of energy

Solution Approach 1:

By merging streams into stream sets, the patent reduces the number of separate garbage collection operations needed. Each stream set can be managed as a single entity, consolidating garbage collection activities and reducing the overall overhead and energy consumption associated with these maintenance operations.

Inventive Principle:

Principle #5Merging (Combining)

3Reliability

If the data storage device treats each stream separately, then stream data integrity is maintained, but write amplification increases

Engineering Contradiction:

Improvestream data integrityVSAvoidwrite amplification

Core Design Contradiction:

ReliabilityVSLoss of energy

Solution Approach 1:

The patent combines streams with similar characteristics into stream sets, allowing write operations to be consolidated. This merging enables the device to reduce write amplification by batching and optimizing write operations across multiple streams while still maintaining the integrity and separation of individual stream data through the stream set structure.

Inventive Principle:

Principle #5Merging (Combining)

4Adaptability or versatility

If the data storage device opens multiple streams for different data lifetimes, then data management flexibility is improved, but device complexity increases

Engineering Contradiction:

Improvedata management flexibilityVSAvoidstream management complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent merges multiple individual stream management tasks into stream set management. By grouping streams with similar performance metrics together, the device reduces the complexity of tracking and managing each stream separately while preserving the flexibility to handle different data lifetimes and access patterns through the stream set abstraction.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent changes the management parameter from individual stream tracking to stream set-based tracking. By monitoring performance metrics at the stream set level rather than individual stream level, the device reduces management complexity while maintaining the ability to adapt to different data access patterns through metric-based classification.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS11853595B2Adaptive multi-stream classification for data storage systems

Publication Date: 2023.12.26 SANDISK TECHNOLOGIES LLC

AI summary

A stream set classification process may be implemented to classify streams opened by a host device on a data storage device. The data storage device may internally classify the streams into different stream classifications using a set of performance metrics. Stream classifications that cause the data storage device to show the greatest gains when compared with a set of baseline performance metrics for the data storage device and/or when compared with other stream classifications, may be selected by the data storage device and/or the host device for subsequent write operations.