Direct-Mapped Flash Storage with OS Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Traditional storage systems face inefficiencies in data management and reliability due to unnecessary write operations and lack of direct control over storage processes, leading to potential data loss and reduced system reliability.

Innovation Solution

Implementing a direct-mapped flash storage system where the operating system initiates and controls storage processes, such as data allocation and garbage collection, without involving lower-level storage controllers, and using non-volatile RAM as a fast-write buffer to improve latency and reliability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If traditional storage controllers manage storage processes, then storage operations can be automated, but system complexity increases and reliability decreases due to multiple layers of abstraction

Engineering Contradiction:
Improvestorage process automationVSAvoidstorage controller complexity
Core Design Contradiction:
Extent of automationVSDevice complexity

Solution Approach 1:

The patent extracts the storage management functions from the storage controller and places them directly in the operating system. The OS initiates and controls storage processes such as data allocation and garbage collection without involving lower-level storage controllers, thereby reducing device complexity while maintaining automation.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The storage system is designed to be self-managing through direct OS control. The OS directly manages data allocation, garbage collection, and other storage processes without requiring intermediate storage controller intervention, simplifying the overall system architecture.

Inventive Principle:
Principle #25Self-service

2Extent of automation

If traditional storage controllers manage data operations, then data management is automated, but data integrity may be compromised due to unnecessary write operations

Engineering Contradiction:
Improvedata management automationVSAvoiddata integrity
Core Design Contradiction:
Extent of automationVSReliability

Solution Approach 1:

By removing the storage controller from the data management chain and having the OS directly control storage operations, the patent eliminates unnecessary intermediate write operations that could compromise data integrity, while preserving automated data management capabilities.

Inventive Principle:
Principle #2Taking out (Extraction)

3Ease of operation

If storage controllers are involved in every storage operation, then fine-grained control is achieved, but system performance decreases due to increased latency

Engineering Contradiction:
Improvestorage control precisionVSAvoidstorage operation speed
Core Design Contradiction:
Ease of operationVSSpeed

Solution Approach 1:

The patent eliminates the storage controller from the data path, allowing the OS to directly control storage operations. This reduces the number of communication hops and latency sources while maintaining precise control over storage processes through direct OS-to-storage-device communication.

Inventive Principle:
Principle #2Taking out (Extraction)

4Reliability

If multiple write operations are performed for data safety, then data reliability is improved, but storage efficiency decreases due to redundant operations

Engineering Contradiction:
Improvedata reliabilityVSAvoidstorage efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

By having the OS directly manage storage operations without storage controller intervention, the system eliminates redundant write operations that were previously necessitated by controller-mediated data paths, thereby improving storage efficiency while maintaining data reliability through direct OS control.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentUS11734097B1Machine learning-based hardware component monitoring
Publication Date: 2023.08.22 PURE STORAGE INC
  • US11734097B1 patent drawing
  • US11734097B1 patent drawing
  • US11734097B1 patent drawing

AI summary

An illustrative method includes identifying, based on an output of a machine learning model that receives data associated with an operation of a hardware component as an input, an anomaly in the data, determining that the anomaly is representative of an issue associated with the hardware component, and performing, based on the determining that the anomaly is representative of the issue associated with the hardware component, a remedial action that affects a performance of the operation of the hardware component.