On-Demand Predictive Analysis with Dynamic Container Execution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The execution of predictive models in data processing systems consumes an undesirable quantity of computing resources, limiting the availability and quality of computer-implemented services by diverting resources from their primary functions.
Innovation Solution
Implementing predictive models within container instances that are dynamically instantiated and terminated based on resource availability and prediction needs, using telemetry data to manage resource allocation efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If predictive models are executed continuously to monitor and predict future system states, then prediction accuracy and system reliability are improved, but computing resource consumption increases
Solution Approach 1:
The patent implements dynamic execution of predictive models based on system state conditions. The system transitions between different execution modes (full prediction, reduced prediction, no prediction) depending on whether the system is in a normal state, degraded state, or recovered state. This dynamic approach ensures high reliability when needed while reducing resource consumption during normal operation.
Solution Approach 2:
The system changes the execution parameters of predictive models based on system state. When the system is in a degraded state, the system increases prediction frequency and computational resources. When in normal state, it reduces prediction frequency. This parameter adaptation resolves the contradiction between reliability and resource consumption.
2Reliability
If predictive models run continuously to provide real-time predictions, then service availability is improved, but computing resources are diverted from primary functions
Solution Approach 1:
The system dynamically adjusts predictive model execution based on service availability status. When services are healthy, prediction execution is reduced or suspended. When services show degradation signs, prediction frequency increases. This dynamic adjustment maintains service availability while optimizing resource allocation to primary functions.
Solution Approach 2:
The system uses its own operational data (telemetry) to automatically trigger predictive model execution without external intervention. The self-monitoring mechanism ensures services are predicted only when necessary, optimizing the balance between availability and resource efficiency.
3Adaptability or versatility
If multiple predictive models are executed to cover different forecasting scenarios, then system adaptability is improved, but device complexity increases
Solution Approach 1:
The patent implements a universal predictive model execution framework that can handle multiple forecasting scenarios through a single system. The framework provides unified management of different predictive models, standardized execution patterns, and centralized state monitoring. This multi-functional approach maintains adaptability while reducing management complexity through consolidation.
Data Source
AI summary
Methods and systems for managing operation of a data processing system that hosts virtual machines that contribute to computer implemented services provided by the data processing system are disclosed. The virtual machines may collect telemetry data on the data processing system to share with a predictive model that is hosted by a container instance. The predictive model may make a prediction for a future state of the data processing system. The prediction may be shared with a management entity module that monitors operation of the data processing system. The management entity module may implement an action, based on the prediction, to manage the operation of the data processing system.