Instrumented Analytic Model Execution Engine for Multi-Cloud Deployment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in efficiently deploying and managing analytic models across different platforms, integrating diverse data streams, and facilitating collaboration between data science and IT teams due to language and infrastructure barriers, leading to slow updates and scalability issues.
Innovation Solution
An analytic model execution engine with instrumentation for granular performance analysis, utilizing a virtualized execution environment (VEE) and containerized design to abstract models and streams, enabling dynamic configuration of microservices and sensors, and a deployment platform for model execution engines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If analytic models are deployed across multiple platforms with diverse data streams, then model versatility and data integration capability are improved, but system complexity and deployment difficulty increase
Solution Approach 1:
The patent introduces an execution engine as an intermediary layer between analytic models and diverse data streams/platforms. This engine provides standardized interfaces and abstractions that mediate the complexity of integrating different data sources while maintaining model versatility, allowing models to be deployed across multiple platforms without increasing overall system complexity.
Solution Approach 2:
The execution engine is designed with universal capabilities to handle multiple data stream types, platforms, and model formats through a single unified architecture. This multi-functional design enables the system to deploy analytic models across diverse platforms while maintaining consistent performance monitoring and data integration, thereby improving versatility without proportionally increasing complexity.
2Measurement precision
If granular instrumentation is added for performance analysis, then measurement precision for metrics and diagnostics is improved, but device complexity and overhead increase
Solution Approach 1:
The instrumentation system is segmented into modular components that can be independently activated based on specific performance analysis needs. Rather than implementing comprehensive instrumentation everywhere, the system divides monitoring functions into discrete, manageable segments that provide precise metrics only where required, reducing overall complexity while maintaining measurement precision for critical performance parameters.
Solution Approach 2:
The instrumentation framework is designed to be dynamic, allowing the level and type of instrumentation to be adjusted based on operational context and analysis requirements. This enables the system to provide granular performance measurement precision when needed while reducing complexity by deactivating or simplifying instrumentation during normal operation, achieving a balance between measurement accuracy and system complexity.
3Productivity
If real-time monitoring and tuning capabilities are implemented, then productivity and optimization speed are improved, but use of energy and computational resources increase
Solution Approach 1:
The real-time monitoring system implements partial instrumentation focused on critical performance parameters rather than comprehensive monitoring of all system activities. This selective approach enables productivity improvements through targeted optimization while reducing computational resource consumption by avoiding excessive monitoring and analysis of non-critical metrics, achieving a balance between optimization speed and energy usage.
Data Source
AI summary
At an interface an analytic model for processing data is received. The analytic model is inspected to determine a language, an action, an input type, and an output type. A virtualized execution environment is generated for an analytic engine that includes executable code to implement the analytic model for processing an input data stream.