Machine-Learning Release Train Stabilization for Continuous Deployment

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing systems face challenges in stabilizing continuous deployment and integration of software updates due to high effort and resource requirements for monitoring deployment and integration, leading to system instability and potential delays or withdrawals of code releases.

Innovation Solution

A system and method for determining the verve and release intensity score (RIS) of code features, stabilizing release trains, and modifying engagement schedules based on these metrics to promote stable code releases, using machine learning for natural language processing of external factors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If continuous deployment and integration of software updates is implemented, then code release frequency and productivity are improved, but system stability deteriorates due to high effort and resource requirements for monitoring

Engineering Contradiction:

Improvecode release frequencyVSAvoidsystem stability

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system implements automated self-monitoring and self-regulation of code releases through machine learning algorithms that automatically assess verve, determine release intensity scores, and stabilize release trains without requiring manual intervention, thereby maintaining high release frequency while ensuring system stability

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system continuously monitors deployment and integration processes, using machine learning to analyze feedback data in real-time, adjust release parameters dynamically, and maintain optimal balance between productivity and stability through closed-loop control

Inventive Principle:

Principle #23Feedback

2Reliability

If high effort and resources are allocated to monitoring deployment and integration, then system reliability is improved, but device complexity and resource requirements worsen

Engineering Contradiction:

Improvecode release stabilityVSAvoidmonitoring system complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The machine learning system performs multiple functions simultaneously - assessing verve, calculating release intensity scores, stabilizing release trains, and modifying engagement schedules - through a single integrated platform, reducing overall system complexity while maintaining high reliability

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system replaces manual monitoring and decision-making processes with automated machine learning algorithms, eliminating the need for complex human-operated monitoring systems while achieving superior reliability through computational analysis

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If manual monitoring and adjustment of release schedules is performed, then measurement precision is improved, but productivity and response speed worsen

Engineering Contradiction:

Improverelease readiness assessment accuracyVSAvoiddeployment speed

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The machine learning system operates continuously to monitor, assess, and adjust release parameters without interruption, maintaining both high measurement precision through constant analysis and high productivity through uninterrupted automated decision-making and deployment

Inventive Principle:

Principle #20Continuity of useful action

Data Source

PatentUS20250291565A1Code agility and visibility via augmented decision-making algorithm

Publication Date: 2025.09.18 KYNDRYL INC

AI summary

A computer-implemented method may include determining, by a computing device, a verve of a feature; determining, by a computing device, a first release intensity score (RIS) of the feature; stabilizing, by a computing device, a release train based on the verve and the RIS of the feature; and modifying, by a computing device, an engagement schedule based on the stabilized release train.