Cloud Instance Scaling via Workload Pattern Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Migrating legacy enterprise applications from on-premise environments to cloud-computing environments often results in increased operational costs due to the need for 24/7 availability, and traditional manual scaling methods are inefficient, leading to resource wastage and potential service delays.
Innovation Solution
Implementing an automatic pattern determination system that processes timeseries data to generate scaling factors for cloud-computing environments, allowing for dynamic resource allocation based on workload patterns, thereby optimizing instance management and resource utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If enterprise applications are deployed in cloud-computing environments with 24/7 availability requirements, then service reliability is improved, but operational costs increase significantly
Solution Approach 1:
The patent implements dynamic scaling of cloud resources based on actual workload demands. The system automatically adjusts the number of running instances according to monitored performance metrics and predicted workload patterns, transitioning from static 24/7 availability to dynamic availability that matches actual service needs, thereby reducing operational costs while maintaining reliability.
Solution Approach 2:
The system changes the operational parameters of cloud instances by adjusting the number of active instances based on workload analysis. It uses historical timeseries data to identify patterns and predicts future workload requirements, then modifies resource allocation parameters accordingly - running more instances during high-demand periods and fewer or no instances during low-demand periods, optimizing the balance between reliability and cost.
2Ease of operation
If manual scaling methods are used for cloud resources, then ease of operation is maintained, but productivity decreases due to resource wastage and service delays
Solution Approach 1:
The patent implements a self-service automated scaling system that monitors workload metrics, analyzes patterns, and automatically adjusts resource allocation without manual intervention. The system uses timeseries analysis and pattern recognition to predict workload demands and autonomously scales instances accordingly, eliminating the inefficiencies of manual scaling while improving resource efficiency and preventing service delays.
Solution Approach 2:
The system establishes a feedback loop where workload performance metrics are continuously monitored and fed back into the scaling decision-making process. The timeseries analysis engine uses historical feedback data to identify patterns and improve future scaling predictions, creating a closed-loop system that continuously optimizes resource allocation based on actual performance outcomes.
3Reliability
If cloud instances are continuously running to ensure availability, then service reliability is improved, but loss of energy increases
Solution Approach 1:
The patent implements periodic scaling actions based on identified workload patterns. Instead of continuous resource allocation, the system determines optimal periods for instance activation and termination by analyzing historical timeseries data. Instances are started and stopped periodically according to predicted workload requirements, reducing resource consumption while maintaining service continuity during actual demand periods.
Data Source
AI summary
Methods, systems, and computer-readable storage media for receiving a set of timeseries, each timeseries in the set of timeseries representing a parameter of execution of the system, pre-processing each timeseries in the set of timeseries to provide a set of pre-processed timeseries, merging timeseries in the set of timeseries to provide a merged timeseries, generating a consolidated timeseries based on the merged timeseries and a periodicity, deriving a pattern based on the consolidated time series, the pattern defining a scaling factor for each period in a timeframe, and executing, by an instance manager, scaling of the system based on the pattern to selectively scale one or more of instances of the system and controllable resources based on scaling factors of the pattern.