Predictive Metrics for Virtualized Deployments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting the infrastructure requirements for virtualized application deployments, such as CRM applications, often rely on insufficient data from previous implementations, leading to inaccurate estimates of the number and type of physical servers needed, especially when deployments differ by industry or scale.
Innovation Solution
A networked computing environment with a prediction service and data extractor that collects operational data from various virtualized deployments, calculates provisioning power units (PPUs), and generates predictive metrics for infrastructure requirements based on industry, application type, and scale, using a linear regression model to estimate vCPU and memory needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing methods use insufficient data from previous implementations, then the prediction process is simpler, but the accuracy of infrastructure estimates deteriorates
Solution Approach 1:
The prediction system segments the infrastructure estimation process into distinct components: data collection from virtualized deployments, calculation of provisioning power units (PPUs), and generation of predictive metrics. This segmentation allows the system to handle complex prediction tasks by breaking them down into manageable stages, each handled by specialized modules that process specific types of data and perform specific calculations.
Solution Approach 2:
The patent introduces provisioning power units (PPUs) as an intermediary metric that bridges the gap between raw operational data and final infrastructure predictions. PPUs serve as a standardized intermediate representation that translates diverse deployment data into a common framework, enabling accurate predictions without requiring direct complex analysis of all input data variables.
2Measurement precision
If the system collects and analyzes data from multiple virtualized deployments, then the accuracy of predictive metrics improves, but the data collection and processing complexity increases
Solution Approach 1:
The prediction system is designed to universally handle data from multiple virtualized deployments across different industries and scales through a single unified framework. The same PPU calculation methodology and predictive metric generation process apply consistently across diverse deployment scenarios, eliminating the need for separate analysis procedures for each deployment type while maintaining high prediction accuracy.
Solution Approach 2:
The system manages data complexity by transforming diverse operational parameters into standardized provisioning power unit measurements. By changing the representation of infrastructure requirements into PPUs, the system can process and compare data from different deployment scales and industries using a consistent parameter framework, thereby reducing processing complexity while maintaining predictive accuracy.
Data Source
AI summary
Various examples are disclosed for generating a prediction of server requirements needed to deploy an application. The application can be deployed in virtualized environment in which virtual machines can execute the application. The predicted server requirements can be generated based upon data from other deployments of the application in other virtualized environments.