Predictive Cloud Database Throttling for DBCPU Overuse
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional throttling methods for cloud-based database resources result in service disruptions and inferior customer experience due to heuristic speed adjustments in throttle queues, leading to DBCPU usage imbalances.
Innovation Solution
A two-stage anomaly mitigation strategy using artificial intelligence and machine learning to predict and regulate DBCPU overuse, first by identifying top contributors and throttling their access, and secondly by dynamically adjusting emission speeds based on ingress data requests.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional throttling methods are used to reduce DBCPU usage, then DBCPU overuse is mitigated, but service availability and customer experience deteriorate due to request timeouts
Solution Approach 1:
The system performs preliminary actions by predicting future DBCPU usage patterns using machine learning models before anomalies occur. By analyzing historical usage data and identifying trends, the system proactively adjusts throttling parameters in advance, preventing DBCPU overuse before it impacts service availability while maintaining optimal data throughput.
Solution Approach 2:
The system implements continuous feedback mechanisms by monitoring real-time DBCPU usage metrics and comparing them against predicted values. Based on this feedback, the throttling parameters are dynamically adjusted to maintain DBCPU usage within acceptable ranges while maximizing service availability and throughput. The feedback loop enables the system to learn from past performance and continuously optimize throttling decisions.
2Productivity
If low emission speed is used in throttle queues, then DBCPU usage decreases, but service quality deteriorates due to request timeouts
Solution Approach 1:
The system applies dynamics by making the throttle queue emission speed variable rather than fixed. Machine learning models continuously predict optimal emission speeds based on current system state, user behavior patterns, and DBCPU capacity. This dynamic adjustment allows the system to maintain high emission speeds when DBCPU capacity is available while reducing speeds proactively when capacity constraints are predicted, thereby maintaining both productivity and service quality.
Solution Approach 2:
The system changes parameters by dynamically adjusting multiple throttling parameters including emission speed, queue size, and priority levels based on predicted DBCPU usage patterns. Rather than using a single fixed parameter, the system modifies multiple parameters in coordination to optimize the balance between DBCPU usage efficiency and service quality, preventing request timeouts while maintaining high throughput.
3Device complexity
If heuristic methods are used for speed adjustment, then implementation is simple, but DBCPU usage balance deteriorates leading to overuse
Solution Approach 1:
The system replaces mechanical heuristic adjustment methods with intelligent machine learning-based control. Instead of relying on simple rules of thumb for speed adjustment, the system employs predictive models that analyze historical data, identify usage patterns, and automatically determine optimal throttling parameters. This substitution of mechanical systems with intelligent systems dramatically improves DBCPU usage balance while maintaining manageable complexity through automated decision-making.
Solution Approach 2:
The system implements self-service by enabling the throttling mechanism to automatically adjust its own parameters based on real-time conditions and learned patterns. The machine learning models continuously train on system performance data and autonomously optimize throttling strategies without requiring manual intervention or complex external control systems. This self-service capability improves DBCPU usage balance while keeping the overall system complexity manageable.
Data Source
AI summary
A computer-implemented method is disclosed for predicting, based on a previous usage of a cloud-based computing resource by a number of users of the cloud-based computing resource, a future usage of the cloud-based computing resource. The method includes predicting, based on the predicted future usage of the cloud-based computing resource, an anomaly event at the cloud-based computing resource. The method also includes implementing a first anomaly mitigation action, based on the prediction of the anomaly event at the cloud-based computing resource and re-evaluating a status of the anomaly event at the cloud-based computing resource after the implementation of the first anomaly mitigation action. The method further includes implementing a second anomaly mitigation action at the cloud-based computing resource, based on the re-evaluation of the status of the anomaly event.