Machine-Learning Workload Mapping for Heterogeneous Servers

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Modern cloud-based data centers with heterogeneous servers face inefficiencies and performance degradation due to not all servers being capable of effectively processing modern workloads, leading to failures and significant adverse impacts on hosted applications.

Innovation Solution

An intelligent workload management system using machine learning algorithms to map and distribute workloads to the most suitable servers based on server configurations and workload types, optimizing resource utilization and minimizing latency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional workload distribution methods are used in heterogeneous computing environments, then system complexity is reduced, but resource utilization efficiency deteriorates and performance degradation occurs

Engineering Contradiction:

Improveresource utilization efficiencyVSAvoidworkload management system complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The system enables workloads to self-select appropriate servers by providing workload types with identifying information about their requirements. Each workload autonomously determines its own placement based on server configuration data and compatibility criteria, eliminating the need for complex centralized scheduling algorithms while achieving optimal resource utilization.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system pre-establishes compatibility relationships between workload types and server configurations by maintaining identifying information about both. This preliminary categorization and matching framework is built in advance, allowing rapid workload distribution decisions without real-time complex computations, thus improving efficiency while managing complexity.

Inventive Principle:

Principle #10Preliminary action

2Reliability

If workloads are distributed to any available server, then system operation is simplified, but performance and reliability deteriorate due to incompatible server-workload pairs

Engineering Contradiction:

Improveworkload processing reliabilityVSAvoidworkload distribution operation

Core Design Contradiction:

ReliabilityVSEase of operation

Solution Approach 1:

The system assigns specific identifying information characteristics to different workload types and server configurations, enabling each workload to be matched with servers that have the specific local qualities (capabilities) required for that workload type. This ensures reliable execution by matching workload-specific requirements with corresponding server capabilities.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The identifying information acts as an intermediary between workloads and servers, providing a standardized interface for compatibility assessment. This intermediary layer simplifies the matching process by translating workload requirements and server capabilities into comparable formats, making the distribution process easier while ensuring reliable pairings.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Adaptability or versatility

If manual workload-to-server mapping is performed, then adaptability to specific server capabilities is improved, but productivity decreases due to time-consuming manual configuration

Engineering Contradiction:

Improveworkload-server matching adaptabilityVSAvoidworkload distribution speed

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The system replaces manual mechanical mapping processes with automated electronic identification and matching systems. Workload types and server configurations are represented by machine-readable identifying information that enables automatic compatibility determination, eliminating manual configuration while maintaining high adaptability to specific server capabilities.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system uses identifying information parameters to represent workload requirements and server capabilities, enabling dynamic and flexible matching based on varying parameters. This parameter-based approach allows the system to adapt to different workload types and server configurations automatically, achieving both adaptability and high productivity through automated parameter comparison.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS12461788B2Intelligent management of workloads in heterogeneous computing environment

Publication Date: 2025.11.04 DELL PROD LP

AI summary

Intelligent workload management techniques in a heterogenous computing environment are disclosed. For example, a method comprises obtaining first identifying information associated with a set of workload types, and obtaining second identifying information associated with a set of server configurations associated with a computing environment, wherein one or more server configurations in the set of server configurations are different than one or more other server configurations in the set of server configurations. The method then maps individual ones of the set of workload types to individual appropriate ones of the set of server configurations based on the obtained first and second identifying information, and causes distribution of one or more workloads to one or more servers in the computing environment, in accordance with the mapping, for execution thereon. At least a portion of the obtaining and mapping steps are performed using one or more machine learning algorithms.