ML-Driven Fleet Package Deployment Automation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Deploying packages, including firmware or low-level system code, to a diverse fleet of hardware components across multiple data centers in a cloud environment is challenging due to the need for safe, secure, and reliable deployment, which is complicated by varying device types, potential impact on performance and power, and the requirement for explicit agreements and thorough validation and monitoring.

Innovation Solution

A method and system that utilize machine learning models to create packages and deployment plans, assessing fleet parameters to specify operations for safe and efficient deployment, including scanning, testing, and monitoring to ensure minimal impact on the fleet, using a processor to automatically generate instructions for deployment and health monitoring.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual deployment methods are used for firmware packages across diverse hardware devices, then deployment control and validation can be performed, but deployment time and complexity increase significantly

Engineering Contradiction:
Improvedeployment safetyVSAvoiddeployment time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system enables self-service deployment where the automated deployment engine independently assesses fleet parameters, generates deployment plans, and executes package deployment across diverse devices without requiring manual intervention for each device, thereby maintaining reliability while reducing deployment time

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system dynamically adjusts deployment parameters based on assessed fleet characteristics, automatically modifying deployment strategies to match specific hardware configurations, power constraints, and performance requirements, enabling efficient and reliable deployment across heterogeneous device fleets

Inventive Principle:
Principle #35Parameter changes

2Reliability

If comprehensive validation and monitoring are performed for each device, then deployment reliability improves, but system complexity and resource requirements increase

Engineering Contradiction:
Improvedeployment reliabilityVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system segments the deployment process into distinct phases (assessment, planning, execution, monitoring) and applies different validation depths to different device groups based on their characteristics, reducing overall system complexity while maintaining comprehensive coverage where needed

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system applies comprehensive validation and monitoring only to critical devices or high-risk deployments, while using streamlined processes for less critical devices, balancing reliability requirements with system complexity management

Inventive Principle:
Principle #16Partial or excessive action

3Productivity

If automated deployment is implemented across diverse hardware devices, then deployment efficiency increases, but ensuring compatibility and safety across all device types becomes more difficult

Engineering Contradiction:
Improvedeployment efficiencyVSAvoiddevice compatibility
Core Design Contradiction:
ProductivityVSAdaptability or versatility

Solution Approach 1:

The system implements a universal deployment engine that can handle multiple device types, hardware configurations, and firmware packages through a single automated platform, assessing fleet parameters to adapt to diverse device requirements while maintaining consistent deployment efficiency across all device categories

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system performs preliminary assessment of fleet parameters and device compatibility before deployment execution, identifying suitable target devices and configuring deployment plans in advance to ensure compatibility across diverse hardware types while maintaining automated efficiency

Inventive Principle:
Principle #10Preliminary action

4Reliability

If thorough assessment of fleet parameters is performed, then deployment safety and appropriateness improve, but deployment preparation time increases

Engineering Contradiction:
Improvedeployment safetyVSAvoidpreparation time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs continuous fleet parameter assessment and maintains updated device profiles, so that when deployment is requested, the assessment work is already partially completed or current, reducing preparation time while maintaining thorough safety evaluation

Inventive Principle:
Principle #20Continuity of useful action

Solution Approach 2:

The system conducts preliminary fleet assessments and maintains readiness information in advance, so that when a deployment request occurs, much of the parameter assessment work has already been performed or can be quickly completed based on pre-collected fleet data

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3953805B1Creating and deploying packages to devices in a fleet based on operations derived from a machine learning model
Publication Date: 2024.12.25 MICROSOFT TECHNOLOGY LICENSING LLC
  • EP3953805B1 patent drawingFigure 1
  • EP3953805B1 patent drawingFigure 2
  • EP3953805B1 patent drawingFigure 3

AI summary

Systems and methods for creating and deploying packages to devices in a fleet based on operations derived from a machine learning model or other automated feedback models are provided. As an example, a method for creating a package, including a payload, for deployment to a set of devices is provided. The method includes receiving a payload, where the payload has an associated set of payload parameters concerning a deployment of the payload to the set of devices. The method further includes using a processor, automatically creating the package for the deployment to the set of devices, where the package comprises instructions for deploying the payload to the set of devices, and where the instructions specify at least one of a plurality of operations derived from a machine learning model based at least on a subset of the associated set of payload parameters.