Deployment Ranking Framework for Software Incident-Cause Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Incident management in large multi-layer service-oriented platforms is challenging due to the complexity of interdependent services and microservices, making it difficult to identify potential causes and solutions for software application incidents, especially with constant code deployments and updates.
Innovation Solution
An automated system using a ranking framework to predict incident-causing deployments by generating semantic similarity, topological distance, and temporal scores for candidate code deployment data objects, and ranking them to recommend mitigating changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated ranking framework is implemented to predict incident-causing deployments, then productivity and efficiency are improved, but device complexity increases
Solution Approach 1:
The system segments the incident analysis process into three distinct scoring components: semantic similarity score (comparing deployment descriptions with incident descriptions), topological distance score (measuring distance in service dependency graph), and temporal score (evaluating time proximity). This segmentation allows complex incident prediction to be broken down into manageable, independent scoring modules that can be computed and combined systematically.
Solution Approach 2:
The patent introduces a learning-to-rank model as an intermediary that synthesizes the three separate scores (semantic similarity, topological distance, temporal) into a unified deployment risk ranking. This intermediary component coordinates the multiple scoring mechanisms and transforms raw scores into actionable incident predictions, managing the complexity of integrating diverse evaluation criteria.
2Measurement precision
If multiple scoring mechanisms are used to improve prediction accuracy, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The system changes parameters by evaluating deployments across three distinct dimensions: semantic parameter (description similarity), topological parameter (service graph distance), and temporal parameter (time proximity). Each dimension provides a different type of measurement that contributes to overall prediction accuracy, allowing the system to capture multiple aspects of deployment-risk relationships simultaneously.
Solution Approach 2:
The patent combines three different scoring mechanisms into a composite evaluation framework. Rather than relying on a single measurement approach, the system integrates semantic similarity scoring, topological distance scoring, and temporal scoring into a unified deployment risk assessment, creating a more robust and accurate prediction system through composition of multiple evaluation methods.
3Loss of time
If manual investigation is replaced with automated ranking, then loss of time is reduced, but measurement precision may worsen
Solution Approach 1:
The system performs self-service by automatically generating deployment risk rankings without requiring manual investigator intervention. The learning-to-rank model autonomously processes deployment data, service dependency graphs, and incident information to produce prioritized predictions, enabling the system to serve its own incident prediction function without external human input for each analysis case.
Solution Approach 2:
The patent incorporates feedback mechanisms where the system learns from historical incident data and deployment outcomes to continuously improve its ranking accuracy. By training the learning-to-rank model on past incidents and their associated deployments, the system refines its scoring parameters and improves measurement precision over time while maintaining automated operation.
Data Source
AI summary
Automated system for predicting software application incident-causing deployments using a ranking framework is provided. A plurality of candidate code deployment data objects for an incident comprising an affected service data object may be identified. For each candidate code deployment data object of the plurality of candidate code deployment data objects a semantic similarity score, a topological distance score, and a temporal score may be generated. The plurality of candidate code deployment data objects may be ranked using a ranking model and based on the semantic similarity score, the topological distance score, and the temporal score for each candidate code deployment data object.