Cloud Instance Prediction System for Cost and Stability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying optimal and stable instances for cloud-based applications are manual, time-consuming, and inefficient, relying on human intervention, which leads to suboptimal resource utilization and increased costs.
Innovation Solution
A system and method that uses a processor to predict a combination of optimal and stable instances by analyzing application configuration, identifying relevant parameters, fetching historical data, calculating stability scores, and recommending a combination of spot and on-demand instances based on cost and performance factors, utilizing machine learning models trained with user feedback.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual trial-and-error method is used to identify optimal instances, then human expertise can be applied, but the process becomes time-consuming and inefficient
Solution Approach 1:
The patent replaces the manual mechanical process of trial-and-error instance selection with an automated machine learning system. The ML model processes application requirements and historical instance data to automatically predict optimal instance combinations, eliminating the need for manual intervention while maintaining or improving selection accuracy.
Solution Approach 2:
The system enables self-service by allowing the ML model to autonomously identify optimal instances without human intervention. The model learns from historical data and automatically makes predictions about instance suitability, making the system self-sufficient in performing the instance selection task.
2Adaptability or versatility
If manual intervention is used for instance selection, then flexibility in decision-making is maintained, but errors due to human intervention increase
Solution Approach 1:
The system incorporates feedback mechanisms where the ML model learns from historical instance performance data and user feedback. This continuous learning process improves the model's accuracy over time while maintaining adaptability to different application requirements, reducing errors while preserving flexibility.
Solution Approach 2:
The system performs preliminary analysis by pre-processing historical instance data and training the ML model in advance. This preparatory work enables the system to quickly and accurately make instance recommendations without manual intervention, reducing errors while maintaining adaptability through pre-learned patterns.
3Loss of energy
If spot instances are used to reduce costs, then resource consumption cost decreases, but stability and reliability of instance execution deteriorates
Solution Approach 1:
The system changes the parameter of instance selection by using ML to predict stability scores for spot instances based on historical data. Instead of treating all spot instances equally, the model identifies specific spot instances with high stability scores, allowing cost reduction while maintaining execution reliability through data-driven parameter optimization.
Solution Approach 2:
The system applies partial action by selectively using spot instances only when the ML model predicts high stability, rather than universally using spot instances for all workloads. This selective approach achieves cost reduction while maintaining reliability by using spot instances only in situations where they are likely to be stable.
4Productivity
If automated system is implemented for instance prediction, then manual effort and errors are reduced, but system complexity increases
Solution Approach 1:
The system achieves universality by designing the ML model to handle multiple types of instances (spot and on-demand) and various application requirements through a single unified framework. This multi-functional approach improves productivity while managing complexity by avoiding the need for separate systems for different instance types.
Solution Approach 2:
The ML model acts as an intermediary between application requirements and instance selection. It mediates the complex processing of historical data, stability predictions, and cost considerations, transforming complex inputs into simple instance recommendations, thereby improving productivity while containing system complexity within the model layer.
Data Source
AI summary
A method and a system for providing a combination of optimal and stable instances are disclosed. The method includes receiving a configuration information of an application for execution; identifying parameters related to the application of user; identifying a set of optimal instances based on the identified parameters; fetching a data of historical spot instance(s) from a host platform; predicting a stability score for each of the optimal spot instances based on at least the data of the historical spot instance(s); predicting an intermediate set of optimal and stable spot instances from the at least one optimal spot instance based on the stability score of the optimal spot instances; and predicting the combination of optimal and stable instances, based on at least on a cost factor and based on at least one of the intermediate set of optimal and stable spot instances, and a set of optimal on-demand instances.