AI Surrogate Model for Cloud Application Multi-Objective Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Managing and optimizing cloud applications with micro-service architectures is challenging due to the complexity of scaling, monitoring, and testing as the number of micro-services increases, and determining optimal trade-offs between key performance indicators like latency and error rates is complicated.
Innovation Solution
A system utilizing an AI platform with an input manager, trial manager, and optimization manager that leverages machine learning-based surrogate function learning models and acquisition functions to conduct adaptive trials, compute a Pareto surface, and select an optimal operating point for cloud applications based on multiple key performance indicators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If micro-service architecture is used to enable scalability and rapid deployment, then business agility and deployment speed are improved, but system complexity and difficulty of monitoring increase
Solution Approach 1:
The patent introduces an AI platform as an intermediary system that manages and optimizes micro-service architectures. This platform provides centralized monitoring, scaling decisions, and performance optimization across multiple micro-services, reducing the operational complexity while maintaining the architectural benefits of micro-services
Solution Approach 2:
The system dynamically adjusts operational parameters of micro-services based on performance metrics and business goals. By automatically tuning parameters such as resource allocation, scaling factors, and configuration settings, the system maintains optimal performance while reducing the manual complexity of managing numerous micro-services
2Reliability
If multiple key performance indicators are optimized simultaneously, then overall application performance is improved, but the complexity of determining optimal trade-offs increases
Solution Approach 1:
The patent implements a feedback-driven optimization system that continuously monitors multiple key performance indicators and uses AI algorithms to determine optimal trade-offs. The system learns from performance data and automatically adjusts parameters to balance competing objectives such as latency, error rates, and resource utilization, simplifying the multi-objective optimization process
Solution Approach 2:
The AI platform performs self-service optimization by autonomously analyzing performance metrics and making scaling decisions without requiring manual intervention. The system automatically identifies optimal operating points and adjusts micro-service configurations to maintain performance across multiple KPIs, reducing the complexity of multi-objective management
3Measurement precision
If adaptive trial execution with machine learning is implemented, then optimization precision is improved, but computational resources and time required increase
Solution Approach 1:
The system performs preliminary actions by pre-training machine learning models with historical performance data before actual optimization trials. This pre-computation allows the AI platform to make faster, more accurate predictions during runtime, reducing the time required for adaptive trial execution while maintaining high optimization precision
Solution Approach 2:
The patent uses machine learning surrogate models that create simplified copies of the complex micro-service system behavior. These surrogate models can rapidly evaluate performance outcomes without requiring full system execution, enabling precise optimization predictions with reduced computational time and resources
Data Source
AI summary
A system, computer program product, and method are provided for orchestrating a multi objective optimization of an application. A set of two or more key performance indicators (KPIs) and one or more parameters associated with the application are received. A machine learning (ML) based surrogate function learning model in combination with an acquisition function is leveraged to conduct one or more adaptive trials. Each trial consists of a specific configuration of the one or more parameters. A pareto surface of the KPIs of the application is computed based on the observations of KPI values from each adaptive trial. The pareto surface is explored and an optimal operating point is selected for the application. The application is then executed at the selected operating point.