Machine Learning Storage Configuration Adjustment

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional storage management approaches in Storage-as-a-Service (STaaS) environments are often reactive, leading to configuration errors and increased costs due to the rarity of low array utilization periods and scheduled outages, which are increasingly less common.

Innovation Solution

The implementation of machine learning techniques to automatically adjust storage system configurations in STaaS environments by obtaining performance-related data, processing it using rule-based analyses, and determining adjustment amounts to optimize storage system configurations without disrupting user workloads.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If conventional storage management approaches are used, then management activities can be performed during low array utilization periods, but configuration errors and increased costs occur due to the rarity of such periods

Engineering Contradiction:

Improvestorage management reliabilityVSAvoidtime for management activities

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs storage management activities in advance by proactively adjusting configurations during low-utilization periods before they are needed, rather than reactively when problems occur. The machine learning model predicts future storage needs and pre-optimizes configurations, ensuring management tasks are completed before utilization peaks occur.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The storage system automatically manages its own configurations through machine learning-driven autonomous adjustment. The system monitors its own performance metrics, identifies optimization opportunities, and self-adjusts configurations without external intervention, enabling continuous self-optimization during low-utilization periods.

Inventive Principle:

Principle #25Self-service

2Productivity

If storage management is performed during low array utilization periods, then user workloads are not affected, but such periods are increasingly less common

Engineering Contradiction:

Improvestorage management efficiencyVSAvoidavailability of low utilization periods

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

The system dynamically adjusts storage configurations based on real-time utilization patterns and workload characteristics. Rather than relying on static low-utilization windows, the machine learning model continuously adapts to changing conditions, identifying and exploiting transient optimization opportunities as they occur, making the system flexible to varying utilization patterns.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes operational parameters such as cache allocation, striping configurations, and redundancy levels based on predicted workload patterns. By dynamically adjusting these parameters in response to utilization changes, the system maintains optimization opportunities even when traditional low-utilization periods disappear.

Inventive Principle:

Principle #35Parameter changes

3Ease of operation

If reactive storage management is used, then management is simpler, but configuration errors and increased costs result

Engineering Contradiction:

Improvestorage management simplicityVSAvoidconfiguration accuracy

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The system implements continuous feedback loops where performance metrics are monitored, analyzed by machine learning models, and used to drive configuration adjustments. This closed-loop approach automatically corrects suboptimal configurations and prevents errors by continuously comparing actual performance against target performance levels, eliminating the need for complex manual tuning.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent replaces manual, human-operated storage management with automated machine learning systems. This substitution eliminates human error in configuration management while maintaining operational simplicity through autonomous decision-making algorithms that automatically optimize storage parameters without requiring user intervention.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS11836365B2Automatically adjusting storage system configurations in a storage-as-a-service environment using machine learning techniques

Publication Date: 2023.12.05 DELL PROD LP

AI summary

Methods, apparatus, and processor-readable storage media for automatically adjusting storage system configurations in a storage-as-a-service environment using machine learning techniques are provided herein. An example computer-implemented method includes obtaining performance-related data for at least one storage system in a storage-as-a-service environment; processing at least a portion of the obtained performance-related data using one or more rule-based analyses; identifying, based at least in part on results of the processing, one or more storage system configurations, of the at least one storage system, requiring adjustment; determining, using at least one machine learning technique, one or more adjustment amounts for the one or more storage system configurations; and automatically adjusting the one or more storage system configurations, within the storage-as-a-service environment, in accordance with the one or more determined adjustment amounts.