Automatic Storage Tiering via Predictive ML Models

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional tiered data storage systems require manual and static storage policies for optimizing data storage tiers, which is inefficient and prone to data access latency, as they lack the ability to automatically adapt to changing access patterns.

Innovation Solution

The implementation of a method for automatic storage tiering that involves receiving characteristics of previous accesses, generating a predictive model using machine learning algorithms, selecting the appropriate storage tier based on future access patterns, and relocating data objects to optimize storage tier alignment, thereby reducing manual intervention and improving performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of energy

If manual storage policies are used to control data tiering, then storage cost can be optimized, but data access latency increases and the system cannot adapt to changing access patterns

Engineering Contradiction:

Improvestorage costVSAvoiddata access latency

Core Design Contradiction:

Loss of energyVSLoss of time

Solution Approach 1:

The system employs machine learning models that automatically analyze access patterns and make autonomous decisions about data tiering, eliminating the need for manual policy configuration. The system self-adjusts to changing access patterns by continuously learning from historical data and automatically relocating files between storage tiers based on predicted future access behavior.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The storage system transitions from static manual policies to dynamic automated decision-making. The machine learning models continuously update their predictions based on new access patterns, allowing the system to adapt in real-time. File relocation decisions are dynamically adjusted based on current system state and predicted future access, optimizing both cost and performance.

Inventive Principle:

Principle #15Dynamics

2Device complexity

If static storage policies are implemented, then system complexity is reduced, but adaptability to changing access patterns deteriorates

Engineering Contradiction:

Improvepolicy management complexityVSAvoidadaptability to access patterns

Core Design Contradiction:

Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent replaces manual mechanical policy configuration with automated machine learning-based decision systems. Instead of administrators manually adjusting storage policies, the system uses algorithms that automatically analyze access patterns and make tiering decisions, substituting human-operated mechanical processes with intelligent automated systems.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system implements continuous feedback loops where access patterns are monitored, analyzed by machine learning models, and used to automatically adjust storage tier assignments. The system learns from past decisions and outcomes, continuously improving its predictions and adaptations to changing access patterns without manual intervention.

Inventive Principle:

Principle #23Feedback

3Speed

If frequent data relocation is performed to optimize storage tiers, then access performance is improved, but system overhead and energy consumption increase

Engineering Contradiction:

Improvedata access speedVSAvoidrelocation energy consumption

Core Design Contradiction:

SpeedVSUse of energy by moving object

Solution Approach 1:

The system uses machine learning models to predict future access patterns before actual access occurs. By anticipating which files will be accessed and when, the system proactively relocates data to appropriate storage tiers in advance, ensuring optimal performance is already in place when access occurs, rather than reacting after access patterns are observed.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system performs selective data relocation based on predicted access patterns rather than moving all data. The machine learning models identify only those files that benefit from relocation, avoiding unnecessary movement of data that would consume energy without providing performance benefits. This partial action approach optimizes the trade-off between relocation overhead and access performance.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS11137926B1Systems and methods for automatic storage tiering

Publication Date: 2021.10.05 ARCTERA US LLC

AI summary

The disclosed computer-implemented method for automatic storage tiering may include (1) receiving characteristics of previous accesses to storage system objects stored in a data storage system including multiple storage tiers, (2) generating, based on the characteristics of previous accesses to the storage system objects, a model that predicts characteristics of future accesses to the storage system objects, (3) selecting, based on the model, a next storage tier of the multiple storage tiers for each of the storage system objects, and (4) relocating at least some of the storage system objects from a current storage tier to the next storage tier selected for each of the at least some of the storage system objects. Various other methods, systems, and computer-readable media are also disclosed.