Distributed Application Parameter Optimization via Utility Curves

The patent performs optimization experimentation in advance in a test execution environment before deploying to production. Candidate sets of application parameter values are tested and evaluated beforehand, so that when the optimized parameters are deployed to the production environment, the actual optimization benefit is achieved without requiring further experimentation that could cause downtime. This preliminary action resolves the contradiction by separating the experimentation phase from the production phase.

The patent creates a copy of the production execution environment as a test execution environment with identical configuration. Optimization experimentation is performed on this copied environment rather than directly on the production system. The test environment accurately replicates production conditions, allowing safe experimentation without risking actual production availability, thus resolving the contradiction between performance improvement and downtime risk.

The patent transforms multiple qualitative stakeholder preferences into quantitative utility curves with measurable parameters. Each stakeholder's preferences are converted into mathematical functions that map application parameter values to utility scores. This parameter transformation allows the system to handle multiple stakeholder requirements simultaneously through computational optimization rather than complex manual decision-making, resolving the contradiction between comprehensiveness and decision complexity.

The patent introduces an optimization component as an intermediary between stakeholders and the distributed application. This component receives preferences from multiple stakeholders, processes them through utility curves and multi-attribute utility rollup operations, and automatically determines optimized parameter values. The intermediary eliminates the need for direct complex negotiations between stakeholders and simplifies the decision-making process into an automated computational task.

The patent replaces manual optimization processes with an automated optimization component that uses computational algorithms. The system automatically performs multi-attribute utility rollup operations, evaluates candidate parameter sets, and determines optimized values without human intervention. This substitution of automated computation for manual analysis resolves the contradiction by making optimization tractable even for complex distributed applications with many services and parameters.

The patent segments the optimization problem into manageable components: the optimization component separates preference processing from parameter evaluation, and the system tests candidate parameter sets independently before deployment. This segmentation allows the complex optimization task to be broken down into discrete, automated steps that can be handled systematically, reducing the perceived difficulty despite the application's complexity.